W "belongs" to ikiwiki and should not be edited directly.
-## [darcs](http://darcs.net/) (not yet included)
+## [[darcs]]
-Support for using darcs as a backend is being worked on by [Thomas
-Schwinge](mailto:tschwinge@gnu.org), although development is on hold curretly.
-There is a patch in [[todo/darcs]].
+Regarding the repository layout: There are two darcs repositories. One is the `srcdir`, the other we'll call `master`.
-### How will it work internally?
+* HTML is generated from `srcdir`.
+* CGI edits happen in `srcdir`.
+* The backend pulls updates from `master` into `srcdir`, i.e. darcs commits should happen to `master`.
+* `master` calls ikiwiki (through a wrapper) in its apply posthook, i.e. `master/_darcs/prefs/defaults` should look like this:
-``Master'' repository R1.
-
-RCS commits from the outside are installed into R1.
-
-HTML is generated from R1. HTML is automatically generated (by using a
-``post-hook'') each time a new change is installed into R1. It follows
-that rcs_update() is not needed.
-
-There is a working copy of R1: R2.
-
-CGI operates on R2. rcs_commit() will push from R2 to R1.
-
-You browse the wiki on R1 and web-edit it on R2. This means for example
-that R2 needs to be updated from R1 if you are going to web-edit a page,
-as the user otherwise might be irritated otherwise...
-
-How do changes get from R1 to R2? Currently only internally in
-rcs\_commit(). Is rcs\_prepedit() suitable?
-
-It follows that the HTML rendering and the CGI handling can be completely
-separated parts in ikiwiki.
-
-What repository should [[RecentChanges]] and History work on? R1?
-
-#### Rationale for doing it differently than in the Subversion case
-
-darcs is a distributed RCS, which means that every checkout of a
-repository is equal to the repository it was checked-out from. There is
-no forced hierarchy.
-
-R1 is nevertheless called the master repository. It's used for
-collecting all the changes and publishing them: on the one hand via the
-rendered HTML and on the other via the standard darcs RCS interface.
-
-R2, the repository the CGI operates on, is just a checkout of R1 and
-doesn't really differ from the other checkouts that people will branch
-off from R1.
-
-(To be continued.)
-
-#### Another possible approach
-
-Here's what I (tuomov) think, would be a “cleaner” approach:
-
- 1. Upon starting to edit, Ikiwiki gets a copy of the page, and `darcs changes --context`.
- This context _and_ the present version of the page are stored in as the “version” of the
- page in a hidden control of the HTML.
- Thus the HTML includes all that is needed to generate a patch wrt. to the state of the
- repository at the time the edit was started. This is of course all that darcs needs.
- 2. Once the user is done with editing, _Ikiwiki generates a patch bundle_ for darcs.
- This should be easy with existing `Text::Diff` or somesuch modules, as the Web edits
- only concern single files. The reason why the old version of the page is stored in
- the HTML (possibly compressed) is that the diff can be generated.
- 3. Now this patch bundle is applied with `darcs apply`, or sent by email for moderation…
- there are many possibilities.
-
-This approach avoids some of the problems of concurrent edits that the previous one may have,
-although there may be conflicts, which may or may not propagate to the displayed web page.
-(Unfortunately there is not an option to `darcs apply` to generate some sort of ‘confliction resolution
-bundle’.) Also, only one repository is needed, as it is never directly modified
-by Ikiwiki.
-
-This approach might be applicable to other distributed VCSs as well, although they're not as oriented
-towards transmitting changes with standalone patch bundles (often by email) as darcs is.
-
-> The mercurial plugin seems to just use one repo and edit it directly - is
-> there some reason that's okay there but not for darcs? I agree with tuomov
-> that having just the one repo would be preferable; the point of a dvcs is
-> that there's no difference between one repo and another. I've got a
-> darcs.pm based on mercurial.pm, that's almost usable... --bma
-
->> IMHO it comes down to whatever works well for a given RCS. Seems like
->> the darcs approach _could_ be done with most any distributed system, but
->> it might be overkill for some (or all?) While there is the incomplete darcs
->> plugin in [[todo/darcs]], if you submit one that's complete, I will
->> probably accept it into ikiwiki.. --[[Joey]]
-
->>> I'd like to help make a robust darcs (2) backend. I also think ikiwiki should use
->>> exactly one darcs repo. I think we can simplify and say conflicting web
->>> edits are not allowed, like most current wiki engines. I don't see that
->>> saving (so much) context in the html is necessary, then.
->>> bma, I would like to see your code. --[[Simon_Michael]]
->>> PS ah, there it is. Let's continue on the [[todo/darcs]] page.
+ apply posthook ikiwrap
+ apply run-posthook
+* The backend pushes CGI edits from `srcdir` back into `master` (triggering the apply hook).
+* The working copies in `srcdir` and `master` should *not* be touched by the user, only by the CGI or darcs, respectively.
## [[Git]]
-Here's Thomas Schwinge unfinished darcs support for ikiwiki.
-
-(Finishing this has been suggested as a [[soc]] project.)
-
-> I haven't been working on this for months and also won't in the near
-> future. Feel free to use what I have done so
-> far and bring it into an usable state! Also, feel free to contact me
-> if there are questions.
-
--- [Thomas Schwinge](mailto:tschwinge@gnu.org)
-
-[[!toggle text="show"]]
-[[!toggleable text="""
- # Support for the darcs rcs, <URL:http://darcs.net/>.
- # Copyright (C) 2006 Thomas Schwinge <tschwinge@gnu.org>
- #
- # This program is free software; you can redistribute it and/or modify it
- # under the terms of the GNU General Public License as published by the
- # Free Software Foundation; either version 2 of the License, or (at your
- # option) any later version.
- #
- # This program is distributed in the hope that it will be useful, but
- # WITHOUT ANY WARRANTY; without even the implied warranty of
- # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- # General Public License for more details.
- #
- # You should have received a copy of the GNU General Public License along
- # with this program; if not, write to the Free Software Foundation, Inc.,
- # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
-
-
- # We're guaranteed to be the only instance of ikiwiki running at a given
- # time. It is essential that only ikiwiki is working on a particular
- # repository. That means one instance of ikiwiki and it also means that
- # you must not `darcs push' into this repository, as this might create
- # race conditions, as I understand it.
-
-
- use warnings;
- use strict;
- use IkiWiki;
-
- package IkiWiki;
-
-
- # Which darcs executable to use.
- my $darcs = ($ENV{DARCS} or 'darcs');
-
-
- # Internal functions.
-
- sub darcs_info ($$$) {
- my $field = shift;
- my $repodir = shift;
- my $file = shift; # Relative to the repodir.
-
- my $child = open(DARCS_CHANGES, "-|");
- if (! $child) {
- exec($darcs, 'changes', '--repo=' . $repodir, '--xml-output', $file) or
- error('failed to run `darcs changes\'');
- }
-
- # Brute force for now. :-/
- while (<DARCS_CHANGES>) {
- last if /^<\/created_as>$/;
- }
- ($_) = <DARCS_CHANGES> =~ /$field=\'([^\']+)/;
- $field eq 'hash' and s/\.gz//; # Strip away the `.gz' from `hash'es.
-
- close(DARCS_CHANGES) or error('`darcs changes\' exited ' . $?);
-
- return $_;
- }
-
-
- # Exported functions.
-
- sub rcs_update () {
- # Not needed.
- }
-
- sub rcs_prepedit ($) {
- # Prepares to edit a file under revision control. Returns a token that
- # must be passed to rcs_commit() when the file is to be commited. For us,
- # this token the hash value of the latest patch that modifies the file,
- # i.e. something like its current revision. If the file is not yet added
- # to the repository, we return TODO: the empty string.
-
- my $file = shift; # Relative to the repodir.
-
- my $hash = darcs_info('hash', $config{srcdir}, $file);
- return defined $hash ? $hash : "";
- }
-
- sub rcs_commit ($$$) {
- # Commit the page. Returns `undef' on success and a version of the page
- # with conflict markers on failure.
-
- my $file = shift; # Relative to the repodir.
- my $message = shift;
- my $rcstoken = shift;
-
- # Compute if the ``revision'' of $file changed.
- my $changed = darcs_info('hash', $config{srcdir}, $file) ne $rcstoken;
-
- # Yes, the following is a bit convoluted.
- if ($changed) {
- # TODO. Invent a better, non-conflicting name.
- rename("$config{srcdir}/$file", "$config{srcdir}/$file.save") or
- error("failed to rename $file to $file.save: $!");
-
- # Roll the repository back to $rcstoken.
-
- # TODO. Can we be sure that no changes are lost? I think that
- # we can, if we make sure that the `darcs push' below will always
- # succeed.
-
- # We need to revert everything as `darcs obliterate' might choke
- # otherwise.
- # TODO: `yes | ...' needed? Doesn't seem so.
- system($darcs, "revert", "--repodir=" . $config{srcdir}, "--all") and
- error("`darcs revert' failed");
- # Remove all patches starting at $rcstoken.
- # TODO. Something like `yes | darcs obliterate ...' seems to be needed.
- system($darcs, "obliterate", "--quiet", "--repodir" . $config{srcdir},
- "--match", "hash " . $rcstoken) and
- error("`darcs obliterate' failed");
- # Restore the $rcstoken one.
- system($darcs, "pull", "--quiet", "--repodir=" . $config{srcdir},
- "--match", "hash " . $rcstoken, "--all") and
- error("`darcs pull' failed");
-
- # We're back at $rcstoken. Re-install the modified file.
- rename("$config{srcdir}/$file.save", "$config{srcdir}/$file") or
- error("failed to rename $file.save to $file: $!");
- }
-
- # Record the changes.
- # TODO: What if $message is empty?
- writefile("$file.log", $config{srcdir}, $message);
- system($darcs, 'record', '--repodir=' . $config{srcdir}, '--all',
- '--logfile=' . "$config{srcdir}/$file.log",
- '--author=' . 'web commit <web-hurd@gnu.org>', $file) and
- error('`darcs record\' failed');
-
- # Update the repository by pulling from the default repository, which is
- # master repository.
- system($darcs, "pull", "--quiet", "--repodir=" . $config{srcdir},
- "--all") and error("`darcs pull' failed\n");
-
- # If this updating yields any conflicts, we'll record them now to resolve
- # them. If nothing is recorded, there are no conflicts.
- $rcstoken = darcs_info('hash', $config{srcdir}, $file);
- # TODO: Use only the first line here, i.e. only the patch name?
- writefile("$file.log", $config{srcdir}, 'resolve conflicts: ' . $message);
- system($darcs, 'record', '--repodir=' . $config{srcdir}, '--all',
- '--logfile=' . "$config{srcdir}/$file.log",
- '--author=' . 'web commit <web-hurd@gnu.org>', $file) and
- error('`darcs record\' failed');
- my $conflicts = darcs_info('hash', $config{srcdir}, $file) ne $rcstoken;
- unlink("$config{srcdir}/$file.log") or
- error("failed to remove `$file.log'");
-
- # Push the changes to the main repository.
- system($darcs, 'push', '--quiet', '--repodir=' . $config{srcdir}, '--all')
- and error('`darcs push\' failed');
- # TODO: darcs send?
-
- if ($conflicts) {
- my $document = readfile("$config{srcdir}/$file");
- # Try to leave everything in a consistent state.
- # TODO: `yes | ...' needed? Doesn't seem so.
- system($darcs, "revert", "--repodir=" . $config{srcdir}, "--all") and
- warn("`darcs revert' failed.\n");
- return $document;
- } else {
- return undef;
- }
- }
-
- sub rcs_add ($) {
- my $file = shift; # Relative to the repodir.
-
- # Intermediate directories will be added automagically.
- system($darcs, 'add', '--quiet', '--repodir=' . $config{srcdir},
- '--boring', $file) and error('`darcs add\' failed');
- }
-
- sub rcs_recentchanges ($) {
- warn('rcs_recentchanges() is not implemented');
- return 'rcs_recentchanges() is not implemented';
- }
-
- sub rcs_notify () {
- warn('rcs_notify() is not implemented');
- }
-
- sub rcs_getctime () {
- warn('rcs_getctime() is not implemented');
- }
-
- 1
-"""]]
-
-This is my ([bma](bma@bmalee.eu)) darcs.pm - it's messy (my Perl isn't up to much) but seems to work. It uses just one repo, like the mercurial plugin (unlike the above version, which AIUI uses two).
-
-`rcs_commit()` uses backticks instead of `system()`, to prevent darcs' output being sent to the browser and mucking with the HTTP headers (`darcs record` has no --quiet option). And `rcs_recentchanges()` uses regexes rather than parsing darcs' XML output.
-
-[[!toggle text="show" id="bma"]]
-[[!toggleable id="bma" text="""
-
- #!/usr/bin/perl
-
- use warnings;
- use strict;
- use IkiWiki;
- use Date::Parse;
- use open qw{:utf8 :std};
-
- package IkiWiki;
-
- sub rcs_update () {
- # Do nothing - there's nowhere to update *from*.
- }
-
- sub rcs_prepedit ($) {
- }
-
- sub rcs_commit ($$$;$$) {
- my ($file, $message, $rcstoken, $user, $ipaddr) = @_;
-
- # $user should probably be a name and an email address, by darcs
- # convention.
- if (defined $user) {
- $user = possibly_foolish_untaint($user);
- }
- elsif (defined $ipaddr) {
- $user = "Anonymous from $ipaddr";
- }
- else {
- $user = "Anonymous";
- }
-
- $message = possibly_foolish_untaint($message);
-
- # BUG: this outputs one line of text, and there's not a -q or --quiet
- # option. Redirecting output to /dev/null works, but I still get the
- # HTTP status and location headers displayed in the browser - is that
- # darcs' fault or ikiwiki's?
- # Doing it in backticks *works*, but I'm sure it could be done better.
- my @cmdline = ("darcs", "record", "--repodir", "$config{srcdir}",
- "-a", "-m", "$message", "--author", "$user", $file);
- `darcs record --repodir "$config{srcdir}" -a -m "$message" --author "$user" $file`; # Return value? Output? Who needs 'em?
- #if (system(@cmdline) != 0) {
- # warn "'@cmdline' failed: $!";
- #}
-
- return undef; # success
-
- sub rcs_add ($) {
- my ($file) = @_;
-
- my @cmdline = ("darcs", "add", "--repodir", "$config{srcdir}", "-a", "-q", "$file");
- if (system(@cmdline) != 0) {
- warn "'@cmdline' failed: $!";
- }
- }
-
- sub rcs_recentchanges ($) {
- # TODO: This is horrible code. It doesn't work perfectly, and uses regexes
- # rather than parsing Darcs' XML output.
- my $num=shift;
- my @ret;
-
- return unless -d "$config{srcdir}/_darcs";
-
- my $changelog = `darcs changes --xml --summary --repodir "$config{srcdir}"`;
- $changelog = join("", split(/\s*\n\s*/, $changelog));
- my @changes = split(/<\/patch>.*?<patch/m, $changelog);
-
-
- foreach my $change (@changes) {
- $change =~ m/hash='(.*?)'/;
- my $rev = $1;
- $change =~ m/author='(.*?)'/;
- my $user = $1."\n";
- my $committype = "web";
- if($user =~ m/</) {
- # Author fields generated by darcs include an email address: look for the "<".
- $committype = "darcs";
- use HTML::Entities;
- $user = decode_entities $user;
- }
- $change =~ m/local_date='(.*?)'/;
- my $when = $1;
- $when=time - str2time($when, 'UTC');
- $change =~ m/<name>(.*?)<\/name>/g;
- my @message = {line => $1};
- foreach my $match ($change =~ m/<comment>(.*?)<\/comment>/gm) {
- push @message, {line => $1};
- }
-
- my @pages;
- foreach my $match ($change =~ m/<.*?_(file|directory)>(.*?)(<(added|removed)_lines.*\/>)*<\/.*?_(file|directory)>/g) {
- # My perl-fu is weak. I'm probably going about this all wrong, anyway.
- push @pages, {page => pagename($match)} if ( -f $config{srcdir}."/".$match || -d $config{srcdir}."/".$match) and not $match =~ m/^$/;
- }
- push @ret, { rev => $rev,
- user => $user,
- committype => $committype,
- when => $when,
- message => [@message],
- pages => [@pages],
- }
- }
- return @ret;
- }
-
- sub rcs_notify () {
- # TODO
- }
-
- sub rcs_getctime ($) {
- error gettext("getctime not implemented");
- }
-
- 1
-
-
-
-"""]]
-
----
-
-Well, here's my version too. It only does getctime -- using a real XML parser, instead of regexp ugliness -- and maybe recentchanges, but that may be bitrotted, or maybe I never finished it, as I only need the getctime. As for actual commits, I have previously voiced my opinion, that this should be done by the plugin generating a patch bundle, and forwarding it to darcs in some way (`darcs apply` or even email to another host, possibly moderated), instead of the hacky direct modification of a working copy. It could also be faster to getctime in a batch. Just reading in all the changes the first time they're needed, might not be a big improvement in many cases, but if we got a batch request from ikiwiki, we could keep reaing the changes until all the files in this batch request have been met. --[[tuomov]]
-
-[[!toggle text="show" id="tuomov"]]
-[[!toggleable id="tuomov" text="""
-<pre>
-#!/usr/bin/perl
-# Stubs for no revision control.
-
-use warnings;
-use strict;
-use IkiWiki;
-
-package IkiWiki;
-
-sub rcs_update () {
-}
-
-sub rcs_prepedit ($) {
- return ""
-}
-
-sub rcs_commit ($$$) {
- return undef # success
-}
-
-sub rcs_add ($) {
-}
-
-sub rcs_recentchanges ($) {
- my $num=shift;
- my @ret;
-
- eval q{use Date::Parse};
- eval q{use XML::Simple};
-
- my $repodir=$config{srcdir};
-
- if (-d "$config{srcdir}/_darcs") {
- my $child = open(LOG, "-|");
- if (! $child) {
- exec("darcs", "changes", "--xml",
- "--repodir", "$repodir",
- "--last", "$num")
- || error("darcs changes failed to run");
- }
- my $data=<LOG>;
- close LOG;
-
- my $log = XMLin($data, ForceArray => 1);
-
- foreach my $patch ($log->{patch}) {
- my $date=$patch->{local_date};
- my $hash=$patch->{hash};
- my $when=concise(ago(time - str2time($date)));
- my @pages;
-
- my $child = open(SUMMARY, "-|");
- if (! $child) {
- exec("darcs", "annotate", "-s", "--xml",
- "--match", "hash: $hash",
- "--repodir", "$repodir")
- || error("darcs annotate failed to run");
- }
- my $data=<SUMMARY>;
- close SUMMARY;
-
- my $summary = XMLin("<lame>$data</lame>", ForceArray => 1);
-
- # TODO: find @pages
-
- push @ret, {
- #rev => $rev,
- user => $patch->{author},
- #committype => $committype,
- when => $when,
- #message => [@message],
- pages => [@pages],
- }; # if @pages;
- return @ret if @ret >= $num;
- }
- }
-
- return @ret;
-}
-
-sub rcs_notify () {
-}
-
-sub rcs_getctime ($) {
- my $file=shift;
-
- eval q{use Date::Parse};
- eval q{use XML::Simple};
- local $/=undef;
-
- # Sigh... doing things the hard way again
- my $repodir=$config{srcdir};
-
- my $filer=substr($file, length($repodir));
- $filer =~ s:^[/]+::;
-
- my $child = open(LOG, "-|");
- if (! $child) {
- exec("darcs", "changes", "--xml", "--reverse",
- "--repodir", "$repodir", "$filer")
- || error("darcs changes $filer failed to run");
- }
-
- my $data=<LOG>;
- close LOG;
-
- my $log = XMLin($data, ForceArray => 1);
-
- my $datestr=$log->{patch}[0]->{local_date};
-
- if (! defined $datestr) {
- warn "failed to get ctime for $filer";
- return 0;
- }
-
- my $date=str2time($datestr);
-
- debug("found ctime ".localtime($date)." for $file");
-
- return $date;
-}
-
-1
-</pre>
-"""]]
-
----
-
-I merged the two versions above and made some fixes; it is recording my web edits in darcs and showing a recent changes page.
-It is in a [darcs repository](http://joyful.com/darcsweb/darcsweb.cgi?r=ikiwiki-darcs), please send patches. --[[Simon_Michael]]
-
-> I'd like to see at least the following fixed before I commit this: --[[Joey]]
-> * Running `darcs record $filename` in backticks is not good (security)
-> The thing to do is to open stdout to /dev/null before execing darcs.
-> * Get `rcs_recentchanges_xml` working, parsing xml with regexps does
-> not seem like a maintenance win.
-> * `rcs_notify` should be removed, it's no longer used.
-> * Some form of conflict handling. Using darcs to attempt to merge
-> the changes is I gusss optional (although every other rcs backend,
-> including svn manages to do this), but it needs to at *least* detect
-> conflicts and return a page with conflict markers for the user to fix
-> the conflict.
-
-I have addressed the recentchanges bit, you can find my hacked up darcs.pm at <http://web.mornfall.net/stuff/web-root/IkiWiki/Rcs/darcs.pm>.
-
-It's got couple of FIXMEs, and a very site-specific filter for recentchanges. Not sure how to do that better though. I will eventually add web commits, probably of my own (and mention it here).
-
----
-
-And here's yet another one, including an updated `ikiwiki-makerepo`. :)
-
<http://khjk.org/~pesco/ikiwiki-darcs/> (now a darcs repo)
> Note that there's a 'darcs' branch in git that I'm keeping a copy of your
> code in. Just in case. :-)
-I've taken all the good stuff from the above and added the missing hooks. The code hasn't seen a lot of testing, so some bugs are likely yet to surface. Also, I'm not experienced with perl and don't know where I should have used the function `possibly_foolish_untaint`.
-
-Regarding the repository layout: There are two darcs repositories. One is the `srcdir`, the other we'll call `master`.
-
- * HTML is generated from `srcdir`.
- * CGI edits happen in `srcdir`.
- * The backend pulls updates from `master` into `srcdir`, i.e. darcs commits should happen to `master`.
- * `master` calls ikiwiki (through a wrapper) in its apply posthook, i.e. `master/_darcs/prefs/defaults` should look like this:
-
- apply posthook ikiwrap
- apply run-posthook
-
- (I'm not sure, should/could it be `ikiwrap --refresh` above?)
- * The backend pushes CGI edits from `srcdir` back into `master` (triggering the apply hook).
- * The working copies in `srcdir` and `master` should *not* be touched by the user, only by the CGI or darcs, respectively.
+I've taken all the good stuff from the above (now deleted --[[Joey]]) and added the missing hooks. The code hasn't seen a lot of testing, so some bugs are likely yet to surface. Also, I'm not experienced with perl and don't know where I should have used the function `possibly_foolish_untaint`.
> Review of this one:
>
> * `rcs_remove` just calls "rm"? Does darcs record notice the file was removed
> and automatically commit the removal? (And why `system("rm")` and not
> `unlink`?)
-> * Is the the darcs info in [[details]] still up-to-date re this version?
+> * Is the the darcs info in [[rcs/details]] still up-to-date re this version?
> --[[Joey]]
> Update:
> this version works. It's similar, but the details differ slightly.
> You could copy my description above to replace it.
>
+>> done --[[Joey]]
+>
> There is still some ironing to do, for instance the current version doesn't allow for
> modifying attachments by re-uploading them via CGI ("darcs add failed"). Am I assuming
> correctly that "adding" a file that's already in the repo should just be a no-op?
sipb-www / ikiwiki.git / commitdiff
