and 
is represented as 
,
where is to be applied first, then A.3
David Roundy
Darcs is a revision control system, along the lines of CVS or arch. That means that it keeps track of various revisions and branches of your project, allows for changes to propogate from one branch to another. Darcs is intended to be an ``advanced'' revision control system. Darcs has two particularly distinctive features which differ from other revision control systems: 1) each copy of the source is a fully functional branch, and 2) underlying darcs is a consistent and powerful theory of patches.
Really, only the first of these three layers is of particular interest to me, so the other two are done as simply as possible. At the database layer, darcs just has an ordered list of patches along with the patches themselves, each stored as an individual file. Darcs' distribution system is strongly inspired by that of arch. Like arch, darcs uses a dumb server, typically apache or just a local or network file system. Unlike arch, darcs currently has no write ability to a remote file system. This means that darcs currently only supports for ``pulling'' of patches from a remote repository to a local one, but not ``pushing'' of patches. While this does simplify matters by eliminating issues of user permissions, it isn't really adequate, as it doesn't address the needs of users who lack a server with a permanent net connection to host their repositories. I do have plans for supporting a push mechanism (which, by the way will be accelerated if I hear there is a demand for such a thing, as I personally have no use for it).
In the last paragraph, I explained revision control systems in terms of three layers. One can also look at them as having two distinct uses. One is to provide a history of previous versions. The other is to keep track of changes that are made to the repository, and to allow these changes to be merged and moved from one repository to another. These two uses are distinct, and almost orthogonal, in the sense that a tool can support one of the two uses optimally while providing no support for the other. Darcs is not intended to maintain a history of versions, although it is possible to kludge together such a revision history, either by making each new patch depend on all previous patches, or by tagging regularly. In a sense, this is what the tag feature is for, but the intention is that tagging will be used only to mark particularly notable versions (e.g. released versions, or perhaps versions that pass a time consuming test suite).
As I understand them (and I certainly may be wrong), previous revision control systems originated with their purpose being to keep track of a history of versions, with the ability to merge changes being added as it was seen that this would be desirable. But the fundamental object remained the versions themselves.
In such a system, a patch (I am using patch here to mean an encapsulated set of changes) is uniquely determined by two trees. Merging changes that are in two trees consists of finding a common parent tree, computing the diffs of each tree with their parent, and then cleverly combining those two diffs and applying the combined diff to the parent tree, possibly at some point in the process allowing human intervention, to allow for fixing up problems in the merge such as conflicts.
Finding this parent tree poses problems. This is where DAGs (Directed Acyclic Graphs) come in. A DAG is a convenient way to assure that two trees (tree here meaning a source tree) have just one closest parent. This means that need keep track of the relationships of the trees, a most imposing task, as the number of trees becomes large! It also may put an artificial constraint on the users, by not allowing them to create cyclic loops of relationships between their versions. Since I don't particularly understand these DAGs (or, for that matter, other revision control systems), I'll leave it here and just trust those who have gone before me that they are difficult.
In the world of darcs, the source tree is not the fundamental object, but rather the patch is the fundamental object. Rather than a patch being defined in terms of the difference between two trees, a tree is defined as the result of applying a given set of patches to an empty tree. Moreover, these patches may be reordered (unless there are dependencies between the patches involved) without changing the tree. Thus there is no need to find a common parent when performing a merge. Or, if you like, their common parent is defined by the set of common patches, and may not correspond to any version in the version history (if we kept track of a history).
One useful consequence of darcs' patch-oriented philosophy is that since a patch need not be uniquely defined by a pair of trees (old and new), we can have several ways of representing the same change, which differ only in how they commute and what the result of merging them is. Of course, creating such a patch will require some sort of user input. This is a Good Thing, since the user creating the patch should be the one forced to think about what they really want to change, rather than the user merging the patch. An example of this is the token replace patch (See Section A.5). This feature make it possible to create a patch, for example, which changes every instance of the variable ``stupidly_named_var'' with ``better_var_name'', while leaving ``other_stupidly_named_var'' untouched. When this patch is merged with any other patch involving the ``stupidly_named_var'', that instance will also be modified to ``better_var_name''. This is in contrast to a more conventional merging method which would not only fail to change new instances of the variable, but would also involves conflicts when merging with any patch that modifies lines containing the variable. By more using additional information about the programmer's intent, darcs is thus able to make the process of changing a variable name the trivial task that it really is, which is really just a trivial search and replace, modulo tokenizing the code appropriately.
The patch formalism discussed in Appendix A is what makes darcs' approach possible. In order for a tree to consist of a set of patches, there must be a deterministic merge of any set patches, regardless of the order in which they must be merged. This requires that one be able to reorder patches. While I don't know that the patches are required to be invertible as well, my implementation certainly requires inveribility. In particular, invertibility is required to make use of Theorem 2, which is used extensively in the manipulation of merges.
To summarize, I believe that darcs has a solid theoretical foundation leagues beyond what anyone else has developed. On the down side, darcs is currently slow, possibly still buggy--certainly still buggy in its entirety, but I mean to say that its core may still be buggy. Moreover it is lacking an abundance of features. However, I believe that the theory behind darcs will be the foundation of the next generation of revision control system.
This chapter should walk you through the steps necesary to build darcs for yourself. There are in general two ways to build darcs. One is for building released versions from tarballs, and the other is to build the latest and greatest darcs, from the darcs repo itself.
Please let me know if you have any problems building darcs, or don't have problems described in this chapter and think there's something obsolete here, so I can keep this page up-to-date.
It is a good idea (but not required) to have a recent version of libcurl installed. If not, you will at least need to have either wget or curl installed if you want to be able to grab repos remotely over normal network protocols (ftp or http). You also might want to have scp available if you want to grab your repos over ssh...
To push patches, you will also need to have a working
/usr/bin/sendmail, which is provided by most mail programs, and is
generally available on linux and BSD systems. It's also there on MacOS X.
However, if you don't have this, it won't stop you from building darcs.
To build the documentation (which isn't really necesary, it's included in html form with the tarballs), you will need to have latex installed, and latex2html if you want to build it in html form.
% sudo gcc_select 3.1to avoid crashes. I have also had to run autoconf once when the configure script failed.
% ./configure % make % make install
There are options to configure that you may want to check out with
% ./configure --help
% darcs get -v http://abridgegame.org/repos/darcsand once you have the darcs repository you can bring it up to date with a
% darcs pull
The repository doesn't hold automatically generated files, which include the configure script and the HTML documentation. The documentation you can do without, but the configure script you have to generate using
% autoconfOf course, this means you'll need to have autoconf installed.
After this the build process is the same as in the tarball case:
% ./configure % make % make install
building_darcs.tex) based
on your experience building darcs...
To do so, just record your changes (which you made in the darcs repository)
% darcs record --no-testmaking sure to give the patch a nice descriptive name. The
--no-test options keeps darcs from trying to run the unit tests,
which can be rather time-consuming. Then you can send the patch to me via
email by
% darcs pushThe darcs repository stores the email address to which patches should be pushed by default.
This chapter will lead you through an example use of darcs, which hopefully will allow you to get started using darcs with your project.
Creating your repository in the first place just involves telling darcs to create the special directory (called _darcs) in your project tree, which will hold the revision information. This is done my simply calling from the root directory of your project:
% cd my_project/ % darcs inittreeThis creates the _darcs directory and populates it with whatever files and directories are needed to describe an empty project. You now need to tell darcs what files and directories in your project should be under revision control. You do this using the command darcs add:
% darcs add *.c Makefile.am configure.inWhen you have added all your files (or at least, think you have), you will want to record your changes. ``Recording'' always includes adding a note as to why the change was made, or what it does. In this case, we'll just note that this is the initial version.
% darcs record --all What is the patch name? Initial revision.Note that since we didn't specify a patch name on the command line we were prompted for one. If the environment variable `EMAIL' isn't set, you will also be prompted for your email address. Each patch that is recorded is given a unique identifier consisting of the patch name, its creator's email address, and the date when it was created.
Now that we have created our repository, make a change to one or more of your files. After making the modification run:
% darcs whatsnewThis should show you the modifications that you just made, in the darcs patch format. If you prefer to see your changes in a different format, read Section 4.4, which describs the whats-new command in detail.
Let's say you have now made a change to your project. The next thing to do is to record a patch. Recording a patch consists of grouping together a set of related changes, and giving them a name. It also tags the patch with the date it was recorded and your email address.
To record a patch simply type:
% darcs recorddarcs will then prompt you with all the changes that you have made that have not yet been recorded, asking you which ones you want to include in the new patch. Finally, darcs will ask you for a name for the patch.
You can now rerun whatsnew, and see that indeed the changes you have recorded are no longer marked as new.
As long as you're running a web browser and making your repo available to
the world, you may as well make it easy for people to see what changes
you've made. You can do this by running make installserver, which
installs the program darcs_cgi at /usr/lib/cgi-bin/darcs. You
also will need to create a cache directory named
/var/cache/darcs_cgi, and make sure the owner of that directory is
the same user that your web browser runs as its cgi scripts as. For me,
this is www-data. Now your friends and enemies should be able to easily
browse your repos by pointing their web browsers at http://your.server.org/cgi-bin/darcs. You can read more about this
interface in Chapter 5.
% darcs pull http://your.server.org/repos/yourprojectDarcs will check to see if you have recorded any changes that aren't in my current repository. If so, it'll prompt me for each one, to see which ones I want to add to my repository. Note that you may see a different series of prompts depending your answers, since sometimes one patch depends on another, so if you answer yes to the first one, you won't be prompted for the second if the first depends on it.
Of course, maybe I don't even have a copy of your repository. In that case I'd want to do a
% darcs get --verbose http://your.server.org/repos/yourprojectwhich gets the whole repo.
The darcs patcher (for a few more details see Chapter 6) is a program that allows you to set up a repository to which you can remotely ``push'' patches. Although this isn't necesary to use darcs, it can be useful if you want to give more than one person write access to a repository, or if you do much of your work using a computer on which it is not convenient to run a web server. The patcher receives patches via cryptographically signed email. For each repository you create on a machine a user is created which determines the email address for that repository. Alas, the patcher is considerably harder to set up than darcs itself, which is why this section of the manual is here.
To install the patcher program, you can either install the ``darcs-server'' package (via debian or whatever), or run ``make installserver''. This hopefully will put the programs in the right places.
To use a pushable repository, you also must install the following programs: sudo, gnupg, a mailer configured to receive mail (e.g. exim, sendmail or postfix), and a web server (usually apache). If you want to be able to browse your repository via the web you must also configure your web server to run cgi scripts and make sure the darcs cgi script was properly installed.
To create a repository, as root run the `darcs-createrepo'. You
will be prompted for the email address of the repository and the location
of an existing copy of the repository. If your desired email is
``myproject@my.url'', this will create a user named ``myproject'' with a
home directory of /var/lib/darcs/repos/myproject. FIXME: I have no
idea if the darcs-createrepo program will even run on any system other than
debian. Success reports would be appreciated (or of course bug reports if
it fails).
The ``myproject'' user will be configured to run the darcs patcher on any
emails it receives. However, the patcher will bounce any emails which
aren't signed by a key in the
/var/lib/darcs/repos/myproject/allowed_keys gpg keyring (which is
empty). To give yourself access to this repository you will need to create
a gpg key. If you don't know about public key cryptography, take a look at
the gnupg manual.
You create your gpg key by running (as your normal user):
% gpg --gen-keyYou will be prompted for your name and email address, among other options. To add your public key to the allowed keys keyring. Of course, you can skip this step if you already have a gpg key you wish to use.
You now need to export the public key so we can tell the patcher about it. You can do this will the following comman (again as your normal user):
% gpg --export "email@address" > /tmp/exported_keyAnd now we can add your key to the
allowed_keys:
(as root)> gpg --keyring /var/lib/darcs/repos/myproject/allowed_keys \
--no-default-keyring --import /tmp/exported_key
You can repeat this process any number of times to authorize multiple users
to push to the repository.
You should now be able to push a patch to the repository by running as your normal user, in a working copy of the repository:
% darcs push --sign http://your.computer/repos/myprojectYou may want to add ``push sign'' to the file
_darcs/prefs/defaults
so that you won't need to type --sign every time you want to
push...
The general format of a darcs command is
% darcs COMMAND OPTIONS ARGUMENTS ...Here COMMAND is a command such as add or record, which of course may have one or more arguments. Options have the form
--option or -o, while arguments vary from command to
command. There are many options which are common to a number of different
commands, which will be summarized here.
Every COMMAND accepts --help as an argument, which tells it to
provide a bit of help. Among other things, this help always provides an
accurate listing of the options available with that command, and is
therefore frequently useful as a supplement to the somewhat spartan help
given in this manual.
% darcs COMMAND --help
Most commands also accept the --verbose option, which tells darcs to
provide additional output. The amount of verbosity varies from command to
command.
Darcs optimizes its operations by keeping track of the modification times
of your files. This dramatically speeds up commands such as
whatsnew and record which would otherwise require reading
every file in the repo and comparing it with a reference version. However,
there are times when this can cause problems, such as when running a series
of darcs commands from a script, in which case often a file will be
modified twice in the same second, which can lead to the second
modification going unnoticed. The solution to such predicaments is the
--ignore-times option, which instructs darcs not to trust the file
modification times, but instead to check each file's contents explicitely.
Several commands need to be able to identify you. Conventionally, you
provide an email address for this purpose. The easiest way to do this is
to define an environment variable EMAIL or DARCS_EMAIL (with
the latter overriding the former). You can also override this using the
--author flag to any command. Alternatively, you could set your
email address on a per-repository basis using the ``defaults'' mechanism
for ``ALL'' commands, as described in Appendix B.
Calling darcs with just ``-help'' as an argument gives a brief summary of what commands are available. Calling darcs with the flag ``-version'' tells you the version of darcs you are using. Similarly calling darcs with only ``-commands'' gives a simple list of available commands. This latter arrangement is primarily intended for the use of command-line autocompletion facilities, as are available in bash.
-v |
--verbose |
give verbose output |
Initialize a new source tree as a darcs repository.
Generally you will only call inittree once for each project you work on, and calling it is just about the first thing you do. Just make sure you are in the main directory of the project, and inittree will set up all the directories and files darcs needs in order to start keeping track of revisions for your project.
inittree actually follows a very simple procedure. It simply creates the directories _darcs, _darcs/current and _darcs/patches, and then creates an empty file, _darcs/inventory. However, it is strongly recommended that you use darcs inittree to do this, as this procedure may change in a future version of darcs.
-v |
--verbose |
give verbose output |
--boring |
don't skip boring files | |
--case-ok |
don't refuse to add files differing only in case | |
-r |
--recursive |
add contents of subdirectories |
Add needs to be called whenever you add a new file or directory to your project. Of course, it also needs to be called when you first create the project, to let darcs know which files should be kept track of.
Darcs will refuse to add a file or directory that differs from an existing one only in case. This is because the HFS+ file system used on under MacOS treats such files as being one and the same.
-v |
--verbose |
give verbose output |
Remove should be called when you want to remove a file from your project, but don't actually want to delete the file. Otherwise just delete the file or directory, and darcs will notice that it has been removed.
-v |
--verbose |
give verbose output |
--ignore-times |
don't trust the file modification times |
Display unrecorded changes in the working directory.
What's-new gives you a view of what changes you've made in your working directory that haven't yet been recorded. The changes are displayed in darcs patch format. Darcs whatsnew will return a non-zero value if there are no changes, which can be useful if you just want to see in a script if anything has been modified. FIXME: If I feel motivated, I may create options for different (possibly more user-friendly) views, such as a unidiff view.
-v |
--verbose |
give verbose output |
--no-test |
don't run the test script |
Check the repository for consistency.
Check verifies that the patches stored in the repository, when successively applied to an empty tree, properly recreate the stored current tree.
If you like, you can configure your repository to be able to run a test suite of some sort. There are two ways you can do this. One is to use ``setpref'' to set the ``test'' value to be a command to run. e.g.
% darcs setpref test "sh configure && make && make test"
Alternatively, you can add to your repo a file called darcs_test.
Darcs will recognize this, and make it executable and run it as a test
script. Typically you would make darcs_test a simple script that
compiles your project and possibly runs a test suite. You should make sure
that your darcs_test returns an error code indicating either success
or failure so darcs will know whether or not the code worked.
If you just want to check the consistency of your repository without
running the test, you can call darcs check with the --no-test
option.
-m |
--patch-name PATCHNAME |
name of patch |
-A |
--author EMAIL |
specify author id |
-v |
--verbose |
give verbose output |
--no-test |
don't run the test script | |
-a |
--all |
answer yes to all patches |
--ask-deps |
ask about dependencies | |
--ignore-times |
don't trust the file modification times |
Record is used to name a set of changes and record the patch to the
repository.
Each patch is given a name, which typically would consist of a brief
description of the changes. This name is later used to describe the patch.
The name must fit on one line (i.e. cannot have any embedded newlines). If
you have more to say, stick it in the log. (FIXME: Currently recording a
patch deletes any pending changes that aren't included in the recorded
patch. This is clearly a bug, but is a bit tedious to fix.)
The patch is also flagged with the author of the change, taken by default
from the DARCS_EMAIL environment variable, and if that doesn't
exist, from the EMAIL environment variable. The date on which the
patch was recorded is also included. Currently there is no provision for
keeping track of when a patch enters a given repository.
Finally, each changeset should have a full log (which may be empty). This
log is for detailed notes which are too lengthy to fit in the name. If you
answer that you do want to create a comment file, darcs will open an editor
so that you can enter the comment in. The choice of editor proceeds as
follows. If one of the $DARCSEDITOR, $VISUAL or
$EDITOR environment variables is defined, its value is used (with
precedence proceeding in the order listed). If not, ``vi'', ``emacs'',
``emacs -nw'' and ``nano'' are tried in that order.
Each patch may depend on any number of previous patches. If you choose to make your patch depend on a previous patch, that patch is required to be applied before your patch can be applied to a repo. This can be used, for example, if a piece of code requires that a function be defined, which has was defined in an earlier patch.
If you want to manually define any dependencies for your patch, you can use
the --ask-deps flag, and darcs will ask you for the patch's
dependencies.
FIXME: Change this code to use the Readline module so arrow keys, etc, will work properly for entering in the text.
If you configure darcs to run a test suite, darcs will run this test on the
recorded repo to make sure it is valid. Darcs first creates a pristine
copy of the source tree (in /tmp), then it runs the test, using its
return value to decide if the record is valid. If it is not valid, the
record will be aborted. This is a handy way to avoid making stupid
mistakes like forgetting to `darcs add' a new file. It also can be
tediously slow, so there is an option (--no-test) to skip the test.
-o |
--repo-name REPONAME |
name of output repository |
-t |
--tag-name TAGNAME |
name of version to pull |
-m |
--patch-name PATCHNAME |
name of patch |
-v |
--verbose |
give verbose output |
Get is used to get a local copy of a repository.
I recommend using the --verbose flag to get, as
this command can take a while, and with no feedback, that can be rather
boring.
If the remote repo and the current directory are in the same filesystem and that filesystem supports hard links, get will create hard links for the patch files, which means that the additional storage space needed will be minimal. This shouldn't cause any problems as darcs usually never modifies a patch file, and if it does it deletes the old file and creates a new file, which should mean that it won't affect the other repo which shares the same patch file.
If you want to get a specific version of a repository, you have a couple of
options. The first is to use the --tag-name option, which allows
you to specify a specific version of the repository by its tag, and get
precisely that version. The argument is a regular expression, and get will
grab the most recent tag which matches that regular expression. The other
option is the --patch-name option, which will allow you to get that
patch (i.e. the most recent one matching the specified regular expression)
and all older patches. Note that when specifying a --patch-name,
you may get a version of your code that has never before been seen, if the
patches have gotten themselves reordered. If you ever want to be able to
precisely reproduce a given version, you need to tag it.
-v |
--verbose |
give verbose output |
-m |
--patch-name PATCHNAME |
name of patch |
-a |
--all |
answer yes to all patches |
--ignore-times |
don't trust the file modification times | |
--no-deps |
don't automatically fulfill dependencies |
Pull is used to bring changes made in another repo into the current repo (that is, the one that is the current directory). Pull allows you to bring over all or some of the patches that are in that repo but not in the current one.
When called without a repository argument, pull will use the most recent repository that was either pushed to or pulled from.
The --patch-name argument can be used to specify a regexp, which
should be of the extended type used by egrep. If this option is
used, only patches which match this regexp (along with their dependencies)
are considered.
If you give a --patch-name argument, darcs will silently pull along
any other patches upon which the patches which match the patch-name depend.
So --patch-name bugfix mean ``pull all the patches with `bugfix' in
their name, along with any patches they require.'' If you really only want
the patches with `bugfix' in their name, you should use the
--no-deps option, which is only useful in combination with
--patch-name, and makes darcs only pull in those matching patches
which have no dependencies (apart from other matching patches).
-v |
--verbose |
give verbose output |
-m |
--patch-name PATCHNAME |
name of patch |
-a |
--all |
answer yes to all patches |
-A |
--author EMAIL |
specify author id |
-t |
--to EMAIL |
specify destination email |
-o |
--output FILE |
specify output filename |
-s |
--sign |
sign the patch with your gpg key |
--edit-description |
edit the patch bundle description |
Push is used to prepare a bundle of patches to send to another repository.
When called without a repository argument, pull will use the most recent repository that was either pushed to or pulled from.
The --output and --email flags determine what darcs does with
the patch bundle after creating it. If you provide an --output
argument, the patch bundle is saved to that file. If you give an
--email argument, the bundle of patches is emailed to that person.
If you don't provide either, darcs will prompt you for an email recipient.
The --patchname argument can be used to specify a regexp, which
should be of the extended type used by egrep. If this option is
used, only patches which match this regexp (along with their dependencies)
are considered for pushing.
If you want to include a description or explanation along with the bundle
of patches, you need to specify the --edit-description flag, which
will cause darcs to open up an editor with which you can compose an email
to go along with your patches.
-V |
--verify PUBRING |
verify that the patch was signed by a key in PUBRING |
-v |
--verbose |
give verbose output |
--ignore-times |
don't trust the file modification times | |
--no-resolve-conflicts |
don't try to resolve conflicts | |
--test |
run the test script |
Apply is used to apply a bundle of patches to this repository. Such a bundle may be created using push.
If you specify the --verify PUBRING option, darcs will check that
the patch was gpg-signed by a key which is in PUBRING.
-v |
--verbose |
give verbose output |
Unrecord is used to undo a single recorded patch. It will prompt you
for which patch to unrecord, and then will undo that patch.
Note that unrecord doesn't affect your working copy
of the tree at all, so if you want to completely undo the change, you'll
also need to darcs revert, or do an unpull.
If you don't revert after unrecording, then the changes made by the unrecorded patches are left in your working tree. If these patches are actually from another repository, interaction (either pushes or pulls) with that repository may be massively slowed down, as darcs tries to cope with the fact that you appear to have made a large number of changes that conflict with those present on the other repository. So if you really want to undo the result of a pull operation, use unpull! Unrecord is primarily intended for when you record a patch, realize it needs just one more change, but would rather not have a separate patch for just that one change.
FIXME: The one major bug remaining is that unrecording a `darcs add' causes that add to be lost, while it really should just be stuck pack in `pending'.
-v |
--verbose |
give verbose output |
Unpull is used to undo a single patch that has been pulled from another repository. It will prompt you for which patch to unpull, and then will undo that patch. Beware that unpull undoes the patch both from the repo records AND from the current working directory, and does NOT check that the patch originated with a pull. In otherwords, you could lose precious code by unpulling!
-v |
--verbose |
give verbose output |
--ignore-times |
don't trust the file modification times | |
-a |
--all |
answer yes to all patches |
Revert is used to undo changes make to the local tree which have not yet been recorded. You will be prompted for which changes you wish to undo. Please be cautious with revert, as there is no way to unrevert. Although I could store an unrevert patch somewhere, if there is demand for such a feature...
-d |
--dist-name DISTNAME |
name of version |
-v |
--verbose |
give verbose output |
Create a distribution tarball.
Dist is a handy tool for implementing a ``make dist'' target in your makefile. It creates a tarball of the recorded edition of your tree. Basically, you will typically use it via a makefile rule such as
dist:
./darcs dist --dist-name darcs-`./darcs --version`
darcs dist then simply creates a clean copy of the source tree,
which it then tars and gzips. If you use programs such as autoconf or
automake, you really should run them on the clean tree before tarring it up
and distributing it. You can do this using the pref value ``predist'',
which is a shell command that is run prior to tarring up the distribution:
% darcs setpref predist "autoconf && automake"
-v |
--verbose |
give verbose output |
Darcs mv needs to be called whenever you want to rename or move a file or directory. Unlike remove, mv actually performs the move itself in your working directory. This is why ``mv'' isn't called ``move'', since it is really almost equivalent to the unix command ``mv''. I could add an equivalent command named ``move'' for those who like vowels.
-v |
--verbose |
give verbose output |
Replace allows you to change a specified token whereever it
occurs in the specified files. Tokens here are defined by a
regexp specifying the characters which are allowed. By default
a token corresponds to a C identifier.
FIXME: Currently there is no way to specify a
different regexp (other than [A-Za-z_0-9]) to be used for a token
replace. Actually, if one of your tokens contains a `-' or `.', you will
then get the ``filename'' regexp, which is [A-Za-z_0-9\-\.].
-m |
--patch-name PATCHNAME |
name of patch |
-A |
--author EMAIL |
specify author id |
-v |
--verbose |
give verbose output |
Tag is used to name a version of the tree. Tag differs from record in that it doesn't record any new changes, and it always depends on all patches residing in the repository when it is tagged. This means that one can later reproduce this version of the repository by calling, for example:
% darcs get --tag-name "darcs 3.14" REPOLOCATION
Each tagged version has a version name. The version is also flagged with the person who tagged it (taken by default from the `DARCS_EMAIL' or `EMAIL' environment variable. The date is also included in the version information.
A tagged version automatically depends on all patches in the repo. This allows you to later reproduce precisely that version. The tag does this by depending on all patches in the repo, except for those which are depended upon by other tags already in the repo. In the common case of a sequential series of tags, this means that the tag depends on all patches since the last tag, plut that tag itself.
-t |
--tag-name TAGNAME |
name of version to pull |
-m |
--patch-name PATCHNAME |
name of patch |
Diff can be used to create a diff between two versions which are in your repository.
Diff calls an external ``diff'' command to do the actual work, and passes any unrecognized flags to this diff command. Thus you can call
% darcs diff -t 0.9.8 -t 0.9.10 -- -uto get a diff in the unified format. Actually, thanks to the wonders of getopt you need the ``
--'' shown above before any arguments to diff.
FIXME: I probably should just bite the bullet and figure out how to get
diff flags passed through properly. Any hints from getopt gurus would be
appreciated.
FIXME: I should allow the user to specify the external diff command. Currently it is hardwired to /usr/bin/diff.
Usage example:
% darcs setpref test "echo I am not really testing anything."
Setpref allows you to set a preferences value in a way
that will propogate to other repositories.
If you just want to set the pref value in your
repository only, you can just edit ``_darcs/prefs/prefs''. Changes
you make in that file will be preserved.
The ``_darcs/prefs/prefs'' holds the only preferences information
that can propogate between repositories via pushes and pulls, and the only
way this happens is when the setprefs command is used. Note that although
prefs settings are included in patches, they are not fully version
controlled. In particular, depending on the order in which a series of
merges is perform, you may end up with a different final prefs
configuration. In practice I don't expect this to be a problem, as the
prefs usually won't be changed very often.
The following values are valid preferences options which can be configured using setpref:
sh ./darcs_test''.
The darcs web interface allows you to conveniently browse the
information stored in a repository. You can view the repo by file and see
the history of that file as various patches were applied, or you can browse
in patch view mode, seeing which files were modified in each patch.
The darcs_cgi cgi script allows you to browse changes made in your
darcs repository via the web. To use it with apache, you can install it
using make installserver, and create a cache directory at
/var/cache/darcs_cgi. This cache directory must be writeable by the
cgi script, which for me means chowning it to the user and group
www-data Finally, you should create a directory named repos
in /var/www, in which you will place symlinks to the repos
themselves. Once all this is done, the user can the browse the repos at
http://your.site/cgi-bin/darcs.
The repos directory is configurable via the configuration file
/etc/darcs/cgi.conf. This file can contain comments (any line
starting with a `#' char) and key value pairs with an equal sign in
between. For example:
# This is an example for cgi.conf # reposdir is the directory containing the repositories reposdir = /var/www/repos # cachedir is a directory writeable by www-data (or whatever user your cgi # scripts run as) which is used to cache the web pages. cachedir = /var/cache/darcs
The page, http://your.site/cgi-bin/darcs, displays a listing of all repos available on the server. From this page, the user can get to any available repository.
Clicking on a given repository will take you to the file view page of that repository, which shows a listing of all the files in the repo. This page also has a link to the patch view page, which is a bit more interesting.
In patch view mode, the web interface displays a listing of all the patches in the repo. Clicking on a patch gives a listing of all files that were in the repo at the time that patch was applied.
Clicking on one of the files shows the file contents, with added lines shown in green, and removed ones in red. To the left of each line is a small `+' and `-'. These are links to the patch which added or removed that line.
The darcs-patcher E-mail-based server allows you to (easily?) set up
a centralized darcs repository to which multiple users can commit patches
without giving those users accounts on the server machine.
darcs-patcher authentication is performed using gnupg signatures.
Users use darcs push --sign to email a signed patch to the
repository. darcs-patcher verifies that the patch is signed by an
authorized user. It then runs the darcs_test, if available, to make
sure the patch doesn't break anything (in which case it would reject it),
and applies the patch to the repo.
To use darcs-patcher, you first need to creat a user for its use
(one user per repository, I'm afraid). In this user's home directory,
create a repository named ``repo''. You have to be a bit careful when
creating this repo since if you just run a darcs get on a local repository
owned by a different user, darcs will use hard links for the patch files,
which may not work since you won't have the right to change them later.
This is a bug in darcs get (FIXME).
You next need to set up a gnupg keyring named allowed_keys in the
.gnupg/ directory containing the gnupg public keys of all users
authorized to commit to the repository. Finally, you create a
.forward file containing the single line
``|/usr/sbin/darcs-patcher'' (adjusted for where you actually
install darcs-patcher), which tells your mailer to pipe any mail
that user receives through darcs-patcher. Note also that darcs
itself must reside in the system path.
A convenient utility called darcs-createrepo is provided to easily
set up darcs-patcher-based repositories. This utility creates a new
user and sets that user up to recieve and validate patches via email.
Because darcs-createrepo must create a new user, it needs to be run
as root. It will prompt you for all the information it needs.
darcs-createrepo creates a repo with no users having write
privileges. Moreover, the allowed_keys file is created as
root-owned, so only root will be able to add allowed users. You probably
will want to chown this file to some administrative user. In any
case, you'll need to add a few users' gpg keys to this file via gpg's
--import or --recv-keys commands.
I think a little background on the author is in order. I am a physicist, and think like a physicist. The proofs and theorems given here are what I would call ``physicist'' proofs and theorems, which is to say that while the proofs may not be rigorous, they are practical, and the theorems are intended to give physical insight. It would be great to have a mathematician work on this, but I am not a mathematician, and don't care for math.
From the beginning of this theory, which originated as the result of a series of email discussions with Tom Lord, I have looked at patches as being analagous to the operators of quantum mechanics. I include in this appendix footnotes explaining the theory of patches in terms of the theory of quantum mechanics. I know that for most people this won't help at all, but many of my friends (and as I write this all three of darcs' users) are physicists, and this will be helpful to them. To nonphysicists, perhaps it will provide some insight into how at least this physicist things.
A patch describes a change to the tree. It could be either a primitive patch (such as a file add/remove, a directory rename, or a hunk replacement within a file), or a compostive patch describing many such changes. Every patch type must satisfy the conditions described in this appendix. The theory of patches is independent of the data which the patches manipulate, which is what makes it both powerful and useful, as it provides a framework upon which one can build a revision control system in a sane manner.
Although in a sense, the defining property of any patch is that it can be applied to a certain tree, and thus make a certain change, this change does not wholly define the patch. A patch is defined by a representation, together with a set of rules for how it behaves (which it has in common with its patch type). The representation of a patch defines what that change that particular patch makes, and must be defined in the context of a specific tree. The theory of patches is a theory of the many ways one can change the representation of a patch to place it in the context of a different tree. The patch itself is not changed, since it describes a single change, which must be the same regardless of its representationA.1.
So how does one define a tree, or the context of a patch? The simplest way to define a tree is as the result of a series of patches applied to the empty treeA.2. Thus, the context of a patch consists of the set of patches that precede it.
The simplest relationship between two patches is that of ``sequential''
patches, which means that the context of the second patch (the one on the
left) consists of the first patch (on the right) plus the context of the
first patch. The composition of two patches (which is also a patch) refers
to the patch which is formed by first applying one and then the other. The
composition of two patches, and is represented as ,
where is to be applied first, then A.3
There is one other very useful relationship that two patches can have,
which is to be parallel patches, which means that the two patches have an
identical context (i.e. their representation applies to identical trees).
This is represented by
. Of course, two patches may also
have no simple relationship to one another. In that case, if you want to
do something with them, you'll have to manipulate them with respect to
other patches until they are either in sequence or in parallel.
The most fundamental and simple property of patches is that they must be
invertible. The inverse of a patch is decribed by: 
. In the
darcs implementation, the inverse is required to be computable from
knowledge of the patch only, without knowledge of its context, but that
(although convenient) is not required by the theory of patches.
is , which is the ``simplest'' patch for
which the composition 
makes no changes to the tree.
The first way (of only two) to change the context of a patch is by commutation, which is the process of changing the order of two sequential patches.
and is represented by
is intended to describe the same change as , with the
only difference being that is applied after 
rather than
before .
' operator should be read as something
like the 
operator in C, indicating that the right hand side performs
identical changes to the left hand side, but the two patches are in
reversed order. When read in this manner, it is clear that commutation
must be a reversible process, and indeed this means that commutation
can fail, and must fail in certain cases. For example, the creation
and deletion of the same file cannot be commuted. When two patches fail to
commute, it is said that the second patch depends on the first, meaning
that it must have the first patch in its context (remembering that the
context of a patch is a set of patches, which is how we represent a tree).
A.4
''.
and is one of two patches,
and , which satisfy the relationship:
The constraint that any two compatible patches (patches which can successfully be applied to the same tree) can be merged is actually quite difficult to apply. The above merge constraints also imply that the result of a series of merges must be independent of the order of the merges. So I'm putting a whole section here for the interested to see what algorithms I use to actually perform the merges (as this is pretty close to being the most difficult part of the code).
The first case is that in which the two merges don't actually conflict, but don't trivially merge either (e.g. hunk patches on the same file, where the line number has to be shifted as they are merged). This kind of merge can actually be very elegantly dealt with using only commutation and inversion.
There is a handy little theorem which is immensely useful when trying to merge two patches.
if and only if
, provided both commutations succeed. If
either commute fails, this theorem does not apply.
on the left, and by 
on the right.
Besides being used in merging, this theorem is also useful in the recursive
commutations of mergers. From Theorem 2, we see that the
merge of and is simply the commutation of with (making sure to do the commutation the right way. Of course, if this
commutation fails, the patches conflict. Moreover, one must check that the
merged result actually commutes with , as the theorem applies only
when both commutations are successful.
Of couse, there are patches that actually conflict, meaning a merge where the two patches truly cannot both be applied (e.g. trying to create a file and a directory with the same name). We deal with this case by creating a special kind of patch to support the merge, which we will call a ``merger''. Basically, a merger is a patch that contains the two patches that conflicted, and instructs darcs basically to resolve the conflict. By construction a merger will satisfy the commutation property (see Definition 3) that characterizes all merges. Moreover the merger's properties are what makes the order of merges unimportant (which is a rather critical property for darcs as a whole).
The job of a merger is basically to undo the two conflicting patches, and then apply some sort of a ``resolution'' of the two instead. In the case of two conflicting hunks, this will look much like what CVS does, where it inserts both versions into the file. In general, of course, the two conflicting patches may both be mergers themselves, in which case the situation is considerably more complicated.
Much of the merger code depends on a routine which recreates from a single merger the entire sequence of patches which led up to that merger (this is, of course, assuming that this is the complicated general case of a merger of mergers of mergers). This ``unwind'' procedure is rather complicated, but absolutely critical to the merger code, as without it we wouldn't even be able to undo the effects of the patches involved in the merger, since we wouldn't know what patches were all involved in it.
Basically, unwind takes a merger such as
M( M(A,B), M(A,M(C,D)))From which it recreates a merge history:
C A M(A,B) M( M(A,B), M(A,M(C,D)))(For the curious, yes I can easily unwind this merger in my head [and on paper can unwind insanely more complex mergers]--that's what comes of working for a few months on an algorithm.) Let's start with a simple unwinding. The merger
M(A,B) simply means that two patches
(A and B) conflicted, and of the two of them A is
first in the history. The last two patches in the unwinding of any merger
are always just this easy. So this unwinds to:
A M(A,B)What about a merger of mergers? How about
M(A,M(C,D)). In this case
we know the two most recent patches are:
A M(A,M(C,D))But obviously the unwinding isn't complete, since we don't yet see where
C and D came from. In this case we take the unwinding of
M(C,D) and drop its latest patch (which is M(C,D) itself) and
place that at the beginning of our patch train:
C A M(A,M(C,D))As we look at
M( M(A,B), M(A,M(C,D))), we consider the unwindings of
each of its subpatches:
C
A A
M(A,B) M(A,M(C,D))
As we did with M(A,M(C,D)), we'll drop the first patch on the
right and insert the first patch on the left. That moves us up to the two
A's. Since these agree, we can use just one of them (they
``should'' agree). That leaves us with the C which goes first.
The catch is that things don't always turn out this easily. There is no
guarantee that the two A's would come out at the same time, and if
they didn't, we'd have to rearrange things until they did. Or if there was
no way to rearrange things so that they would agree, we have to go on to
plan B, which I will explain now.
Consider the case of M( M(A,B), M(C,D)). We can easily unwind the
two subpatches
A C M(A,B) M(C,D)Now we need to reconcile the
A and C. How do we do this?
Well, as usual, the solution is to use the most wonderful
Theorem 2. In this case we have to use it in the reverse of
how we used it when merging, since we know that A and C could
either one be the last patch applied before M(A,B) or
M(C,D). So we can find C' using
C' A M(A,B) M( M(A,B), M(C,D) )There is a bit more complexity to the unwinding process (mostly having to do with cases where you have deeper nesting), but I think the general principles that are followed are pretty much included in the above discussion.
There are a couple of simple constraints on the routine which determines how to resolve two conflicting patches (which is called `glump'). These must be satisfied in order that the result of a series of merges is always independent of their order. Firstly, the output of glump cannot change when the order of the two conflicting patches is switched. If it did, then commuting the merger could change the resulting patch, which would be bad. Secondly, the result of the merge of three (or more) conflicting patches cannot depend on the order in which the merges are performed.
The conflict resolution code (glump) begins by ``unravelling'' the merger into a set of sequences of patches. Each sequence of patches corresponds to one non-conflicted patch that got merged together with the others. The result of the unravelling of a series of merges must obviously be independent of the order in which those merges are performed. This unravelling code (which uses the unwind code mentioned above) uses probably the second most complicated algorithm. Fortunately, if we can successfully unravel the merger, almost any function of the unravelled merger satisfies the two constraints mentioned above that the conflict resolution code must satisfy.
It can sometimes be handy to have a canonical representation of a given patch. We achieve this by defining a canonical form for each patch type, and a function ``canonize'' which takes a patch and puts it into canonical form. This routine is used by the diff function to create an optimal patch (based on an LCS algorithm) from a simple hunk describing the old and new version of a file. Note that canonization may fail, if the patch is internally inconsistent.
A simpler, faster (and more generally useful) cousin of canonize is the coalescing function. This takes two sequential patches, and tries to turn them into one patch. This function is used to deal with ``split'' patches, which are created when the commutation of a primitive patch can only be represented by a composite patch. In this case the resulting composite patch must return to the original primitive patch when the commutation is reversed, which a split patch accomplishes by trying to coalesce its contents each time it is commuted.
There is another handy function, which primarily affects file patches (although it can also affect other patches, such as rename patches or dir add/remove patches), which is the submerge-in-directory function. This function changes the patch to act on a patch within a subdirectory rather than in the current directory, and is useful when performing the recursive diff.
Hunks are an example of a complex filepatch. A hunk is a set of lines of a text file to be replaced by a different set of lines. Either of these sets may be empty, which would mean a deletion or insertion of lines. The hunk is the simplest patch that has a commuting pattern in which the commuted patches differ from the originals (rather than simple success or failure). This makes commuting or merging two hunks a tad tedious. Hunks, of course, can be coalesced if they have any overlap. Note that coalesce code doesn't check if the two patches are conflicting. If you are coalescing two conflicting hunks, you've already got a bug somewhere.
One of the most important pieces of code is the canonization of a hunk,
which is where the ``diff'' algorithm is performed. This algorithm begins
with chopping off the identical beginnings and endings of the old and new
hunks. This isn't strictly necesary, but is a good idea, since this
process is 
, while the primary diff algorithm is something
considerably more painful than that... actually the head would be dealt
with all right, but with more space complexity. I think it's more
efficient to just chop the head and tail off first.
Although most filepatches will be hunks, darcs is clever enough to support
other types of changes as well. A ``token replace'' patch replaces all
instances of a given token with some other version. A token, here, is
defined by a regular expression, which must be of the simple [a-z...] type,
indicating which characters are allowed in a token, with all other
characters acting as delimiters. For example, a C identifier would be a
token with the flag [A-Za-z_0-9].
What makes the token replace patch special is the fact that a token replace can be merged with almost any ordinary hunk, giving exactly what you would want. For example, you might want to change the patch type TokReplace to TokenReplace (if you decided that saving two characters of space was stupid). If you did this using hunks, it would modify every line where TokReplace occurred, and quite likely provoke a conflict with another patch modifying those lines. On the other hand, if you did is using a token replace patch, the only change that it could conflict with would be if someone else had used the token ``TokenReplace'' in their patch rather than TokReplace--and that actually would be a real conflict!
Of course, in order to store our patches in a file, we'll have to save them as some sort of strings. The convention is that each patch string will end with a newline, but on parsing we skip any amount of whitespace between patches.
{
<put patches here> (indented two)
}
( <put patches here> (indented two) )
hunk FILE LINE# -LINE ... +LINE ...
Replace a token with a new token. Note that this format means that the white space must not be allowed within a token. If you know of a practical application of whitespace within a token, let me know and I may change this.
replace FILENAME [REGEX] OLD NEW
Modify a binary file
binary FILENAME oldhex *HEXHEXHEX ... newhex *HEXHEXHEX ...
addfile filename
rmfile filename
move oldname newname
changepref prefname oldval newval
adddir filename
rmdir filename
merger MERGERVERSION <first patch> <second patch>
Named patches are diplayed as a `patch id' which is in square brackets, followed by a patch. Optionally, after the patch id (but before the patch itself) can come a list of dependencies surrounded by angle brackets. Each dependency consists of a patch id.
A repository consists of a working directory, which has within it a
directory called _darcs. There must also be subdirectories named
current and patches. current contains the version of
the tree which has been recorded. patches contains the actual
patches which are in the repository.
Within _darcs is the inventory file, which lists all the
patches that are in the repo. Moreover, it also gives the order of the
representation of the patches as they are stored. Given a source of
patches, i.e. any other set of repositories which have between them all the
patches contained in a given repo, that repo can be reproduced based on
only the information in the inventory file. Under those
circumstances, the order of the patches specified in the inventory
file would be unimportant, as this order is only needed to provide context
for the interperetation of the stored patches in this repository.
There is a very special patch which may be stored in patches which
is called `pending'. This patch describes any changes which have not yet
been recorded, and cannot be determined by a simple diff. For example file
additions or renames are placed in pending until they are recorded.
Similarly, token replaces are stored in pending until they are recorded.
The _darcs directory will also typically contain a prefs
directory. This directory exists simply to hold user configuration
settings specific to this repository. This directory may be freely
deleted, and its contents are intended to be modifiable by the user,
although a mistake in such a modification may cause darcs to behave
strangely or cause your cat to cough up a hairball.
Default values for darcs commands can be configured on a per-repository
basis by editing (and possibly creating) the _darcs/prefs/defaults
file. Each line of this file has the following form:
COMMAND FLAG VALUEwhere
COMMAND is either the name of the command to which the default
applies, or ALL to indicate that the default applies to all commands
accepting that flag. The FLAG term is the name of the long argument
option without the ``--'', i.e. verbose rather than
--verbose. Finally, the VALUE option can be omitted if the
flag is one such as verbose that doesn't involve a value.
For example, if your system clock is bizarre, you could instruct darcs to
always ignore the file modification times by adding the following line to
your _darcs/prefs/defaults file. (Note that this would have to be
done for each repository!)
ALL ignore-times
If you never want to run a test when recording to a particular repository (but still want to do so when running check on that repo), and like to name all your patches ``Stupid patch'', you could use the following:
record no-test record patch-name Stupid patch
_darcs/prefs/repos file contains a list of repositories you have
pulled from or pushed to, and is used for autocompletion of pull and push
commands in bash. Feel free to delete any lines from this list that might
get in there, or to delete the file as a whole.
_darcs/prefs/boring file may contain a list of regular
expressions describing files, such as object files, that you do not expect
to add to your project. As an example, the boring file that I use with
my darcs repository is:
\.hi$ \.o$ ^\.[^/] ^_ ~$ (^|/)CVS($|/)The last line is actually irrelevant because I don't use CVS, but if I did it would be handy. You may want to have the boring file under version control. To do this you can use darcs setpref to set the value ``boringfile'' to the name of your desired boring file (e.g. ``darcs setpref boringfile ./.boring'', where the .boring is a file that has been darcs added to your repository). You could also use this to configure your repo to work with a global boring file, something like ``/etc/darcs/boring''.
_darcs/prefs/binaries file may contain a list of regular
expressions describing files that should be treated as binary files rather
than text files. You probably will want to have the binaries file under
version control. To do this you can use darcs setpref to set the value
``binariesfile'' to the name of your desired binaries file (e.g. ``darcs
setpref binariesfile ./.binaries'', where .binaries is a file that has been
darcs added to your repository).
Copyright © 1989, 1991 Free Software Foundation, Inc.
59 Temple Place - Suite 330, Boston, MA 02111-1307, USA
Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.
The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundation's software and to any other program whose authors commit to using it. (Some other Free Software Foundation software is covered by the GNU Library General Public License instead.) You can apply it to your programs, too.
When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things.
To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it.
For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights.
We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software.
Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations.
Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all.
The precise terms and conditions for copying, distribution and modification follow.
Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does.
You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee.
These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it.
Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Program.
In addition, mere aggregation of another work not based on the Program with the Program (or with a work based on the Program) on a volume of a storage or distribution medium does not bring the other work under the scope of this License.
The source code for a work means the preferred form of the work for making modifications to it. For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.
If distribution of executable or object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place counts as distribution of the source code, even though third parties are not compelled to copy the source along with the object code.
If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply and the section as a whole is intended to apply in other circumstances.
It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system, which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice.
This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License.
Each version is given a distinguishing version number. If the Program specifies a version number of this License which applies to it and ``any later version'', you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.
This document was generated using the LaTeX2HTML translator Version 2K.1beta (1.48)
Copyright © 1993, 1994, 1995, 1996,
Nikos Drakos,
Computer Based Learning Unit, University of Leeds.
Copyright © 1997, 1998, 1999,
Ross Moore,
Mathematics Department, Macquarie University, Sydney.
The command line arguments were:
latex2html -split 0 -external_file darcs -prefix big -no_auto_link -local_icons -dir manual bigpage.tex
The translation was initiated by David Roundy on 2003-07-29
A.3
So I'm sticking with my hand waving formalism.