]> git.vanrenterghem.biz Git - git.ikiwiki.info.git/log
git.ikiwiki.info.git
7 years agoUpdate l10n
Simon McVittie [Sun, 1 Oct 2017 15:32:51 +0000 (16:32 +0100)]
Update l10n

Signed-off-by: Simon McVittie <smcv@debian.org>
7 years agot/git-untrusted.t: Fix redundant declaration
Simon McVittie [Sun, 1 Oct 2017 15:24:48 +0000 (16:24 +0100)]
t/git-untrusted.t: Fix redundant declaration

Signed-off-by: Simon McVittie <smcv@debian.org>
7 years agodebian: Declare compliance with Debian Policy 4.1.1
Simon McVittie [Sun, 1 Oct 2017 12:55:48 +0000 (13:55 +0100)]
debian: Declare compliance with Debian Policy 4.1.1

Signed-off-by: Simon McVittie <smcv@debian.org>
7 years agoUpdate changelog
Simon McVittie [Sun, 1 Oct 2017 12:23:04 +0000 (13:23 +0100)]
Update changelog

Signed-off-by: Simon McVittie <smcv@debian.org>
7 years agot/git-untrusted.t: New test case for untrusted pushes
Simon McVittie [Sun, 1 Oct 2017 12:17:53 +0000 (13:17 +0100)]
t/git-untrusted.t: New test case for untrusted pushes

This also exercises the typical centralized git repository workflow,
where changes flow from a non-bare clone (for example on a laptop)
to a centralized bare repository, then from the centralized bare
repository to a non-bare clone that is ikiwiki's srcdir.

Signed-off-by: Simon McVittie <smcv@debian.org>
7 years agoPass some git environment variables to the test-receive hook
Simon McVittie [Sun, 1 Oct 2017 12:16:03 +0000 (13:16 +0100)]
Pass some git environment variables to the test-receive hook

Since git 2.11, git has stored the proposed push in a "quarantine
area" until it is accepted by the pre-receive hook, and passed
extra environment variables to the pre-receive hook so that it can
read objects from the quarantine area.

This fixes untrusted push on modern git versions.

Signed-off-by: Simon McVittie <smcv@debian.org>
7 years agoIkiWiki::Receive: Avoid using asprintf
Simon McVittie [Sat, 30 Sep 2017 16:14:34 +0000 (17:14 +0100)]
IkiWiki::Receive: Avoid using asprintf

On GNU/Linux, it isn't declared in stdio.h unless we define
_GNU_SOURCE, which we don't; using the implicit declaration risks
crashes on platforms where sizeof(pointer) != sizeof(int). On other
platforms it isn't guaranteed to exist at all.

Signed-off-by: Simon McVittie <smcv@debian.org>
7 years agoUpdate changelog and close bug
Simon McVittie [Thu, 28 Sep 2017 10:21:21 +0000 (11:21 +0100)]
Update changelog and close bug

7 years agoMerge remote-tracking branch 'tails/img-determinism'
Simon McVittie [Thu, 28 Sep 2017 10:19:27 +0000 (11:19 +0100)]
Merge remote-tracking branch 'tails/img-determinism'

7 years agoUpdate changelog
Simon McVittie [Thu, 28 Sep 2017 10:18:26 +0000 (11:18 +0100)]
Update changelog

7 years agoFix build with po4a 0.52.
Amitai Schleier [Wed, 20 Sep 2017 02:48:06 +0000 (22:48 -0400)]
Fix build with po4a 0.52.

7 years ago(no commit message)
fairusurped@12113b58e7c4c69149ded64c82f268e9fa14ad88 [Thu, 7 Sep 2017 13:55:41 +0000 (09:55 -0400)]

7 years ago(no commit message)
azzamsa [Wed, 6 Sep 2017 07:13:24 +0000 (03:13 -0400)]

7 years ago(no commit message)
azzamsa [Wed, 6 Sep 2017 07:13:07 +0000 (03:13 -0400)]

7 years agoAdded a comment: my issue solved
azzamsa [Wed, 6 Sep 2017 07:09:23 +0000 (03:09 -0400)]
Added a comment: my issue solved

7 years agoReport bug + merge request: image resize is not deterministic.
intrigeri [Fri, 1 Sep 2017 19:38:30 +0000 (15:38 -0400)]
Report bug + merge request: image resize is not deterministic.

7 years agoimg: resize images in a deterministic manner.
intrigeri [Fri, 1 Sep 2017 19:26:26 +0000 (19:26 +0000)]
img: resize images in a deterministic manner.

7 years agot/img.t: test determinism of PNG resizing.
intrigeri [Fri, 1 Sep 2017 19:16:27 +0000 (19:16 +0000)]
t/img.t: test determinism of PNG resizing.

7 years agoAdded a comment: Reposted question on unix.sx
anna19 [Mon, 28 Aug 2017 15:49:19 +0000 (11:49 -0400)]
Added a comment: Reposted question on unix.sx

7 years agoemailauth, passwordauth: Avoid leaving cgisess_* files in the system temp directory.
Joey Hess [Wed, 23 Aug 2017 17:13:23 +0000 (13:13 -0400)]
emailauth, passwordauth: Avoid leaving cgisess_* files in the system temp directory.

Due to the use/abuse of CGI::Session to generate a token for the login
process, a new session database was created for each login, and left behind
afterwards. While each file is small, with many logings this could bloat
the size of /tmp significantly. Fixed by making CGI::Session write to
/dev/null, since there does not seem to be a way to entirely prevent the
writing.

This commit was sponsored by Henrik Riomar on Patreon.

7 years agoremoved
Keeh [Mon, 21 Aug 2017 20:02:23 +0000 (16:02 -0400)]
removed

7 years ago(no commit message)
Keeh [Mon, 21 Aug 2017 14:28:51 +0000 (10:28 -0400)]

7 years ago(no commit message)
Keeh [Mon, 21 Aug 2017 14:20:33 +0000 (10:20 -0400)]

7 years agoClarify how to use sorting
tbm@179cf05cd1bd36f352bd555cee25923d4216668f [Sat, 19 Aug 2017 13:23:20 +0000 (09:23 -0400)]
Clarify how to use sorting

7 years ago(no commit message)
dgsb [Wed, 9 Aug 2017 06:07:37 +0000 (02:07 -0400)]

7 years ago(no commit message)
dgsb [Tue, 8 Aug 2017 20:37:01 +0000 (16:37 -0400)]

7 years ago(no commit message)
vpelcak@b216e425210695d731d2673167c7dd45e5e9b1c9 [Mon, 7 Aug 2017 06:49:07 +0000 (02:49 -0400)]

7 years agoAdded a comment
smcv [Sun, 23 Jul 2017 18:02:30 +0000 (14:02 -0400)]
Added a comment

7 years agoAdded a comment
smcv [Sun, 23 Jul 2017 17:52:11 +0000 (13:52 -0400)]
Added a comment

7 years agoAdded a comment
smcv [Sun, 23 Jul 2017 17:47:55 +0000 (13:47 -0400)]
Added a comment

7 years agoAdded a comment
smcv [Sun, 23 Jul 2017 17:46:37 +0000 (13:46 -0400)]
Added a comment

7 years agocore: Don't decode the result of strftime if already tagged as UTF-8
Simon McVittie [Sun, 23 Jul 2017 14:56:24 +0000 (15:56 +0100)]
core: Don't decode the result of strftime if already tagged as UTF-8

It wasn't in old Perls, but might be in Perl >= 5.21.1 due to commit
https://perl5.git.perl.org/perl.git/commit/9717af6 (Closes: #869240)

7 years agoAdded a comment: I have the same issue
azzamsa [Fri, 21 Jul 2017 08:07:39 +0000 (04:07 -0400)]
Added a comment: I have the same issue

7 years agoQ. How to truncate blog posts?
ankit [Tue, 18 Jul 2017 13:07:48 +0000 (09:07 -0400)]
Q. How to truncate blog posts?

7 years agohtmlscrubber: Add support for the video tag's loop and muted attributes.
Joey Hess [Tue, 11 Jul 2017 19:51:44 +0000 (15:51 -0400)]
htmlscrubber: Add support for the video tag's loop and muted attributes.

Those were not in the original html5 spec, but have been added in the
whatwg html living standard and have wide browser support.

This commit was sponsored by John Peloquin on Patreon.

7 years agoanswer question, with reference.
DavidCary [Wed, 5 Jul 2017 17:51:19 +0000 (13:51 -0400)]
answer question, with reference.

7 years agoAdded a comment
openmedi [Sun, 25 Jun 2017 16:12:20 +0000 (12:12 -0400)]
Added a comment

7 years agoremoved
smcv [Fri, 23 Jun 2017 14:28:02 +0000 (10:28 -0400)]
removed

7 years ago(no commit message)
test [Thu, 22 Jun 2017 18:01:30 +0000 (14:01 -0400)]

7 years agorequest more information
Simon McVittie [Thu, 22 Jun 2017 14:37:19 +0000 (15:37 +0100)]
request more information

7 years agot/img.t: Give better diagnostics if we can't load an image
Simon McVittie [Thu, 22 Jun 2017 14:15:41 +0000 (15:15 +0100)]
t/img.t: Give better diagnostics if we can't load an image

7 years agoadd bug report originally emailed to me by Peter Simons
Joey Hess [Thu, 22 Jun 2017 13:55:27 +0000 (09:55 -0400)]
add bug report originally emailed to me by Peter Simons

7 years agoAnnounce 3.20170622
Simon McVittie [Thu, 22 Jun 2017 09:53:23 +0000 (10:53 +0100)]
Announce 3.20170622

7 years agodebian/changelog: Add missing credit for toc fix
Simon McVittie [Thu, 22 Jun 2017 09:53:17 +0000 (10:53 +0100)]
debian/changelog: Add missing credit for toc fix

7 years ago3.20170622 3.20170622 debian/3.20170622
Simon McVittie [Thu, 22 Jun 2017 08:25:25 +0000 (09:25 +0100)]
3.20170622

7 years agodebian: Declare compliance with Debian Policy 4.0.0
Simon McVittie [Thu, 22 Jun 2017 08:24:48 +0000 (09:24 +0100)]
debian: Declare compliance with Debian Policy 4.0.0

7 years agodebian: Use preferred https URL for Format of debian/copyright
Simon McVittie [Thu, 22 Jun 2017 08:24:22 +0000 (09:24 +0100)]
debian: Use preferred https URL for Format of debian/copyright

7 years agodebian: Document more past changes
Simon McVittie [Thu, 22 Jun 2017 08:19:53 +0000 (09:19 +0100)]
debian: Document more past changes

7 years agometa: Specifically document [[!meta foo:bar="baz"]] as not working
Simon McVittie [Thu, 22 Jun 2017 07:59:20 +0000 (08:59 +0100)]
meta: Specifically document [[!meta foo:bar="baz"]] as not working

7 years ago(no commit message)
j@d945f5982c686dda5ab7bc2ef45e09d388233fad [Tue, 20 Jun 2017 23:03:02 +0000 (19:03 -0400)]

7 years agoosm: Convert savestate hook into a changes hook
Simon McVittie [Tue, 20 Jun 2017 22:36:21 +0000 (23:36 +0100)]
osm: Convert savestate hook into a changes hook

savestate is not the right place to write wiki content, and in particular
this breaks websetup if osm's dependencies are not installed, even if
the osm plugin is not actually enabled. (Closes: #719913)

This is not a full solution: it should be possible to render the PoI files
for only the maps that changed, from the format, changes or rendered
hook. However, getting that right would require more understanding of
this plugin, and this version is enough to not break websetup. This
version is the closest correct hook to the one where this previously
took place.

7 years ago(no commit message)
alicef [Mon, 12 Jun 2017 21:14:22 +0000 (17:14 -0400)]

7 years agoEnsure repo gets picked up by gitremotes script
https://tylercipriani.com/ [Fri, 2 Jun 2017 12:55:00 +0000 (08:55 -0400)]
Ensure repo gets picked up by gitremotes script

7 years agoAdd jsonfeed patch
https://tylercipriani.com/ [Thu, 1 Jun 2017 23:26:28 +0000 (19:26 -0400)]
Add jsonfeed patch

7 years agoAdd thcipriani repository
https://tylercipriani.com/ [Thu, 1 Jun 2017 23:17:04 +0000 (19:17 -0400)]
Add thcipriani repository

7 years agoAdd my user page
https://tylercipriani.com/ [Thu, 1 Jun 2017 23:15:33 +0000 (19:15 -0400)]
Add my user page

7 years agocurrent headinganchors does not damage headings' attributes, although it does not...
smcv [Thu, 1 Jun 2017 14:03:51 +0000 (10:03 -0400)]
current headinganchors does not damage headings' attributes, although it does not act on those headings

7 years ago(no commit message)
smcv [Thu, 1 Jun 2017 13:59:36 +0000 (09:59 -0400)]

7 years agoresolved
smcv [Thu, 1 Jun 2017 13:48:10 +0000 (09:48 -0400)]
resolved

7 years agoresponse
anarcat [Thu, 1 Jun 2017 13:14:23 +0000 (09:14 -0400)]
response

7 years agoresponse
anarcat [Thu, 1 Jun 2017 13:02:26 +0000 (09:02 -0400)]
response

7 years ago(no commit message)
https://me.yahoo.com/a/GwQv.Tw.p_ux8p4eLf9CkcwYsQ--#2fdeb [Sat, 27 May 2017 02:25:07 +0000 (22:25 -0400)]

7 years agoAdded a comment: Please do not patch out the symlink check
smcv [Fri, 26 May 2017 06:20:23 +0000 (02:20 -0400)]
Added a comment: Please do not patch out the symlink check

7 years agoAdded a comment: git-annex support
mail@b2ae8518bb6eba14728f86acae7830e4c2d9943d [Thu, 25 May 2017 14:43:20 +0000 (10:43 -0400)]
Added a comment: git-annex support

7 years ago(no commit message)
openmedi [Thu, 25 May 2017 11:30:47 +0000 (07:30 -0400)]

7 years agoAdded a comment: I suggest asking macOS/brew people
smcv [Mon, 22 May 2017 11:02:44 +0000 (07:02 -0400)]
Added a comment: I suggest asking macOS/brew people

7 years ago(no commit message)
qazwsx [Sun, 21 May 2017 23:23:36 +0000 (19:23 -0400)]

7 years ago(no commit message)
qazwsx [Sun, 21 May 2017 23:22:54 +0000 (19:22 -0400)]

7 years agoAdded a comment
openmedi [Fri, 19 May 2017 15:32:18 +0000 (11:32 -0400)]
Added a comment

7 years agoSuggested syntax does work, and has a test
smcv [Fri, 19 May 2017 13:57:28 +0000 (09:57 -0400)]
Suggested syntax does work, and has a test

7 years agoDocument the special case for [[!meta name=foo content=bar]]
smcv [Fri, 19 May 2017 13:50:52 +0000 (09:50 -0400)]
Document the special case for [[!meta name=foo content=bar]]

7 years agoit is (meant to be) possible, just not with that syntax
smcv [Fri, 19 May 2017 13:43:08 +0000 (09:43 -0400)]
it is (meant to be) possible, just not with that syntax

7 years ago(no commit message)
fmarier [Thu, 18 May 2017 17:33:44 +0000 (13:33 -0400)]

7 years agolong out of date
bma@d2ddf927e0bde7166ad151d794bee58589bedb40 [Tue, 16 May 2017 12:59:37 +0000 (08:59 -0400)]
long out of date

7 years agocolor, toc: Fix `make test`
Simon McVittie [Tue, 16 May 2017 10:33:23 +0000 (11:33 +0100)]
color, toc: Fix `make test`

7 years agocolor: Use markup for the preserved CSS, not character data
Simon McVittie [Tue, 16 May 2017 10:29:30 +0000 (11:29 +0100)]
color: Use markup for the preserved CSS, not character data

This still smuggles it past the sanitize step, but avoids having
other plugins that want to capture text content without markup
(notably toc) see the CSS as if it was text content.

7 years agocolor: Add a unit test
Simon McVittie [Tue, 16 May 2017 10:11:01 +0000 (11:11 +0100)]
color: Add a unit test

7 years agowe should prefer existing IDs and only act as a fallback
smcv [Tue, 16 May 2017 09:38:02 +0000 (05:38 -0400)]
we should prefer existing IDs and only act as a fallback

7 years agocross-reference i18nheadinganchors
smcv [Tue, 16 May 2017 09:26:25 +0000 (05:26 -0400)]
cross-reference i18nheadinganchors

7 years agocorrect ID syntax
smcv [Tue, 16 May 2017 09:17:57 +0000 (05:17 -0400)]
correct ID syntax

7 years agobrowsers and specifications support more Unicode than we give them credit for
smcv [Tue, 16 May 2017 09:17:00 +0000 (05:17 -0400)]
browsers and specifications support more Unicode than we give them credit for

7 years agoclose
smcv [Tue, 16 May 2017 08:27:56 +0000 (04:27 -0400)]
close

7 years agoAdd a simple unit test for [[!toc]]
Simon McVittie [Tue, 16 May 2017 07:52:39 +0000 (08:52 +0100)]
Add a simple unit test for [[!toc]]

7 years agoPrune dead links
Simon McVittie [Tue, 16 May 2017 07:36:59 +0000 (08:36 +0100)]
Prune dead links

7 years agoReinstate a git repo that has come back
Simon McVittie [Tue, 16 May 2017 07:36:48 +0000 (08:36 +0100)]
Reinstate a git repo that has come back

7 years agouse heading identifiers in TOC links
Antoine Beaupré [Wed, 12 Apr 2017 19:42:42 +0000 (15:42 -0400)]
use heading identifiers in TOC links

reasoning: if headings have identifiers, they are probably more useful
anchors than the automatically generated anchors we build in the toc
plugin. this can happen if, for example, you use the `multimarkdown`
plugin, which inserts `id` tags for every header it encounters. this
also leverages the `headinganchors` plugin nicely.

keeps backwards-compatibility with old toc-generated #indexXhY
anchors.

7 years agoAdded a comment
smcv [Tue, 16 May 2017 07:29:33 +0000 (03:29 -0400)]
Added a comment

7 years agoClarify documentation
Simon McVittie [Tue, 16 May 2017 07:28:04 +0000 (08:28 +0100)]
Clarify documentation

7 years agoAdd a test-case for Markdown options
Simon McVittie [Tue, 16 May 2017 07:25:53 +0000 (08:25 +0100)]
Add a test-case for Markdown options

7 years agomdwn: Don't enable alphabetically labelled ordered lists by default
Simon McVittie [Tue, 16 May 2017 06:58:12 +0000 (07:58 +0100)]
mdwn: Don't enable alphabetically labelled ordered lists by default

This avoids misinterpreting initials ("C. S. Lewis was an author"),
the abbreviation for Monsieur ("M. Descartes was a philosopher") and
German page numbering ("S. 42") as ordered lists if they happen to
begin a line.

This only affects the default Discount implementation: Text::Markdown
and Text::MultiMarkdown do not have this feature anyway. A new
mdwn_alpha_list option can be used to restore the old interpretation.

7 years agoAdded a comment
qazwsx [Mon, 15 May 2017 06:19:37 +0000 (02:19 -0400)]
Added a comment

7 years agomdwn: Enable footnotes by default when using Discount
Simon McVittie [Sun, 14 May 2017 17:16:53 +0000 (18:16 +0100)]
mdwn: Enable footnotes by default when using Discount

A new mdwn_footnotes option can be used to disable footnotes in
MultiMarkdown and Discount.

7 years agomdwn: Don't mangle <style> into <elyts> under some circumstances
Simon McVittie [Sun, 14 May 2017 16:45:55 +0000 (17:45 +0100)]
mdwn: Don't mangle <style> into <elyts> under some circumstances

We can ask libdiscount not to elide <style> blocks, which means we
don't have to work around them.

7 years agomdwn: Make flags passed to libdiscount explicit
Simon McVittie [Sun, 14 May 2017 16:34:12 +0000 (17:34 +0100)]
mdwn: Make flags passed to libdiscount explicit

The Perl binding defaults to MKD_NOHEADER|MKD_NOPANTS anyway, but
making them explicit means we can use other flags of our choice,
and makes it easier to justify why those flags are appropriate.

7 years agocgierror: When the CGI fails, print the error to stderr, not "Died"
Simon McVittie [Sun, 14 May 2017 14:06:01 +0000 (15:06 +0100)]
cgierror: When the CGI fails, print the error to stderr, not "Died"

$@ could be clobbered by the "exception handler", and in practice
it seems that it is. This can be seen on stderr of t/git-cgi.t.

7 years agoDefend against empty session names
Simon McVittie [Sun, 14 May 2017 13:52:43 +0000 (14:52 +0100)]
Defend against empty session names

If misconfiguration has resulted in an empty session name, treat the
session as having not signed in.

7 years agohttpauth: If REMOTE_USER is empty, behave as though it was unset
Simon McVittie [Sun, 14 May 2017 13:44:43 +0000 (14:44 +0100)]
httpauth: If REMOTE_USER is empty, behave as though it was unset

A frequently cut-and-pasted HTTP basic authentication configuration
for nginx sets it to the empty string when not authenticated, which
is not useful.

7 years agohttpauth: Recommend if_not_empty parameter for REMOTE_USER
Simon McVittie [Sun, 14 May 2017 13:45:30 +0000 (14:45 +0100)]
httpauth: Recommend if_not_empty parameter for REMOTE_USER

This is untested, but should hopefully avoid the failure mode
described in [[bugs/Anon_edit_caused_lock_out_on_entire_site_]].

7 years agot/git-cgi.t: Wait 1 second before doing a revert that should succeed
Simon McVittie [Sun, 14 May 2017 14:09:29 +0000 (15:09 +0100)]
t/git-cgi.t: Wait 1 second before doing a revert that should succeed

This hopefully fixes a race condition in which the test failed
around 6% of the time.

If we don't wait, the mtime (which is rounded down to 1 second precision
in the APIs we use) will not necessarily change, so the update will not
necessarily cause the page to be refreshed.

Bug-Debian: https://bugs.debian.org/862494

7 years agocomplete last paragraph
smcv [Sun, 14 May 2017 12:31:49 +0000 (08:31 -0400)]
complete last paragraph