Parabola Issue Tracker: Issueshttps://labs.parabola.nu/https://labs.parabola.nu/favicon.ico?15367742552014-03-19T12:38:08ZParabola Issue Tracker
Redmine pbot-ng - Bug #512 (open): Sometimes his nick is wronghttps://labs.parabola.nu/issues/5122014-03-19T12:38:08Zxylonjoseph@xylon.me.uk
<p>Like right now he think his name is `pbot' but it's `pbot_'.</p> Servers - Bug #508 (open): [ParabolaWeb] Integrate with the wiki to translate news posts.https://labs.parabola.nu/issues/5082014-03-05T23:56:31Zlukeshulukeshu@parabola.nuServers - Bug #507 (open): [ParabolaWeb] Get package "Maintainer" (and "Contributor") information...https://labs.parabola.nu/issues/5072014-03-05T23:55:27Zlukeshulukeshu@parabola.nu
<p>Instead of having it managed in both SQL manually, and in the ABS tree.</p> Servers - Bug #506 (open): [ParabolaWeb] Compare >2 architectures at the same time.https://labs.parabola.nu/issues/5062014-03-05T23:54:13Zlukeshulukeshu@parabola.nu
<p>Currently, it supports comparing only any 2 architectures.</p> Servers - Bug #468 (open): [ParabolaWeb] /packages/differences/ with an unrecognized architecture...https://labs.parabola.nu/issues/4682014-01-21T17:03:42Zlukeshulukeshu@parabola.nu
<p>Doing a query on <a class="external" href="https://parabolagnulinux.org/packages/differences/">https://parabolagnulinux.org/packages/differences/</a> with an unrecognized architecture currently returns a generic HTTP 500 error page. It should return a 4XX range error (either 404 or 410... perhaps 404, but I'd think abou it), along with a message saying what's up.</p>
<p>I say this because GoogleBot, and occaisionally random other users, periodically try to load a comparision of i586 & i686; i586 having been deleted.</p> Servers - Bug #453 (open): [labs] mail notifications have implicit destinationhttps://labs.parabola.nu/issues/4532013-12-29T13:27:11Zmtjmmtjm@mtjm.eu
<p>Mailman puts all notification emails sent to the dev list in its moderation queue with the reason "Message has implicit destination".</p>
<p>Disabling "Blind carbon copy recipients (bcc)" solves this, but it leaks emails of all subscribers in the mail sent to the list. Is it ok?</p>
<p>Can it be fixed to send a separate mail with the user in the To header?</p>
<p>Or is it a mail server issue?</p> pbot-ng - Bug #409 (open): He can't handle egyptian urlshttps://labs.parabola.nu/issues/4092013-09-13T23:46:23Zxylonjoseph@xylon.me.uk
<p>If you say a URL like:</p>
<p><a class="external" href="http://موقع.وزارة-الأتصالات.مصر/">http://موقع.وزارة-الأتصالات.مصر/</a></p>
<p>He seemingly just ignores it</p> pbot-ng - Bug #405 (open): When he gets page titles he doesn't handle unicodehttps://labs.parabola.nu/issues/4052013-09-03T02:11:21Zxylonjoseph@xylon.me.uk
<p>For example: <a class="external" href="http://www.tudou.com/">http://www.tudou.com/</a></p> Documentation - Bug #339 (open): GPT and syslinux - steps after updating with pacman -Syuhttps://labs.parabola.nu/issues/3392013-05-24T13:57:00Zfchmmrinfo@voipplanet.co.uk
<p>Hi,</p>
<p>I updated my Parabola system with pacman -Syu, which installed a new linux-libre version.</p>
<p>I use GPT partitioning and SYSLINUX bootloader, and after rebooting further to updating, I found that I could not boot into my Parabola system, with the SYSLINUX menu instead giving me the following:</p>
<p>failed to load libcom32.c32<br />failed to load COM32 file vesamenu.c32</p>
<p>Since I could not therefore boot into my system, I attempted a fix via the Live CD, and was successful.</p>
<p>Basically you have to mount your installation in the Live CD and do this (after chrooting into it with eg arch-chroot /mnt):</p>
<p>syslinux-install_update -iam</p>
<p>Relevant data:<br /><a class="external" href="http://pastebin.com/W6Ke8cCD">http://pastebin.com/W6Ke8cCD</a><br /><a class="external" href="http://pastebin.com/xYyei60q">http://pastebin.com/xYyei60q</a></p> Packages - Bug #334 (open): [maven] Uses network during build() to download makedependshttps://labs.parabola.nu/issues/3342013-05-16T18:29:44Zlukeshulukeshu@parabola.nu
<p>This is an updated-version of <<a class="external" href="https://labs.parabola.nu/issues/38&gt;;">https://labs.parabola.nu/issues/38&gt;;</a> the process/details have changed.</p>
<p>I'll try to explain this as best I can so that someone without any prior knowledge can understand.</p>
<a name="About-maven-as-a-package-manager"></a>
<h1 >About maven as a package manager<a href="#About-maven-as-a-package-manager" class="wiki-anchor">¶</a></h1>
<p>Apache maven, in addition to being a build-system is also a primitive package manager. Packaging software that uses maven for it's build-system can be a real pain, because it will try to download pre-compiled dependencies during the build; which is against our packaging policies.</p>
<p>About the maven package manager:</p>
<ul>
<li>A package is called an "artifact" </li>
<li>An artifact has a `groupId`, a name, and a version.</li>
<li>An artifact consists of a `.pom` XML file and (usually) a compiled `.jar` Java file.</li>
<li>Artifacts are installed to `/usr/share/maven/repository/$groupId/$artifact/$version/$artifact-$ver.{pom,jar}` (by default, the location can be configured)</li>
</ul>
<p>Most of the packages in [java] install themselves as an artifact, and symlink the jar file into `/usr/share/java`.</p>
<a name="About-mavens-build-processPKGBUILD"></a>
<h1 >About maven's build process/PKGBUILD<a href="#About-mavens-build-processPKGBUILD" class="wiki-anchor">¶</a></h1>
<p>Maven itself uses a bootstrap version of maven to build itself, and in the process downloads some 560 artifacts. I am confident that these are all free software, so the issue isn't "urgent", but it does violate our policies.</p>
<p>I have set up the maven PKGBUILD to record what artifacts it downloads to `$srcdir/downloaded-artifacts.txt` (I have attached a copy of this for convenience); what we need to do is package all of them and add them to `makedepends=()`; I have already done this with `java-antlr2` (2.7.7).</p>
<a name="About-version-proliferation"></a>
<h1 >About version proliferation.<a href="#About-version-proliferation" class="wiki-anchor">¶</a></h1>
<p>Most of the artifacts downloaded are not directly dependencies of maven, but dependencies of dependencies. This sucks because Java+maven developers tend to over-specify the version that they are dependent on, so all the named versions will be downloaded. For example, it downloads 1.0, 1.0.3, 1.0.4, 1.1, and 1.1.1 of java-commons-logging, each needed by a different artifact.</p>
<p>We should fix this, which requires more in-depth knowledge of the system.</p>
<p>Of course, sometimes multiple major versions are OK, we have java-asm2 and java-asm3. But what's not OK is that asm 3.3.1 is packaged, but it downloads asm 3.1. Also not OK is that antlr 2.7.7 is packaged, but it wants both 2.7.2 and 2.7.7.</p>
<a name="Why-we-include-maven-even-though-it-violates-our-policies"></a>
<h1 >Why we include maven, even though it violates our policies.<a href="#Why-we-include-maven-even-though-it-violates-our-policies" class="wiki-anchor">¶</a></h1>
<p>Right now, most java packages from Arch violate our policies. We are working to fix this. We explicitly allow maven, because it is important in building other packages.</p> Packages - Bug #270 (open): [pacman] misleading error message when cache contains a signed packag...https://labs.parabola.nu/issues/2702012-12-03T09:33:41Zmtjmmtjm@mtjm.eu
<pre>17:24 < fauno> ==> ERROR Package community/rxvt-unicode-9.15-3-mips64el.pkg.tar.xz already exists in another repository
[...]
17:27 < fauno> i'll get this behavior on pacman
17:27 < fauno> community.db will say rxvt-unicode needs to update, but it won't download your version because mine is cached
17:28 < fauno> then it will fail to upgrade because the signature doesn't verify the package
17:28 < fauno> instead of telling you the package is probably corrupt because the checksums don't match like before
17:29 < mtjm> easy to remove it from cache
[...]
17:31 < fauno> mtjm: yes, but the message is misleading
17:31 < fauno> it suggest having a security issue rather than a corrupt package</pre>
<p>Expected behaviour: handle invalid signatures just like invalid checksums, i.e. remove the cached package and error.</p> Servers - Bug #202 (open): [wiki] i18n template shouldn't be displayed with action=render (plain ...https://labs.parabola.nu/issues/2022012-10-12T22:27:36Zalfplayeralfplayer@mailoo.org
<p>ArchISO (iso builder) uses this URL to get the Installation Guide, but the URL doesn't work as it does in ArchWiki or Wikipedia because action=render doesn't work as it should.</p>
<p><code>https://wiki.parabolagnulinux.org/Installation_Guide?action=render</code></p>
<p>This URL works:</p>
<p><code>https://wiki.parabolagnulinux.org/index.php?title=Installation_Guide&action=render</code></p>
<p>The issue with this URL is that the MediaWiki template <a class="external" href="https://wiki.parabolagnulinux.org/Template:I18n">https://wiki.parabolagnulinux.org/Template:I18n</a> appears at the top (it shouldn't).</p>
<p>The template does not appear using action=render on ArchWiki, which uses "interlanguage links" (<a class="external" href="https://wiki.archlinux.org/index.php/Help:I18n#Interlanguage_links">https://wiki.archlinux.org/index.php/Help:I18n#Interlanguage_links</a>) and not the template (thanks ovruni for pointing this out).</p>
<p>Can action=render be fixed to avoid displaying the template, as in ArchWiki?</p> Servers - Bug #173 (open): [labs] Too much data is needed to registerhttps://labs.parabola.nu/issues/1732012-08-09T13:22:03Zmtjmmtjm@mtjm.eu
<p>Another solution would be to accept anonymous reports and comments.</p>
<p>Users reporting problems on #parabola wrote that password, real name and email address shouldn't be required. The first field is clearly useless with OpenID (which we probably don't recommend enough).</p> Servers - Bug #56 (open): [labs] Support tagging issueshttps://labs.parabola.nu/issues/562012-03-29T18:13:00Zmtjmmtjm@mtjm.eu
<p>From #parabola:<br /><pre>
16:54 < mtjm> fauno: does labs.parabola.nu support tagging of issues?
16:55 < mtjm> so e.g. the how to help page could have a link to a list of all open issues which need MIPS assembly knowledge to fix
16:56 < fauno> mmm maybe with a plugin
16:56 < mtjm> Ruby or C might be a better case
16:57 < fauno> https://github.com/friflaj/redmine_tagging
</pre></p> Packages - Bug #34 (open): Change packages showing Arch bug report URL or calling the distro Archhttps://labs.parabola.nu/issues/342012-03-04T14:00:27Zmtjmmtjm@mtjm.eu
<p>The changes are partially done in abslibre-mips64el, but this is not specific to mips64el. Probably adding an issue tracker for packages should be done before.</p>