#archlinux-ports | Logs for 2017-11-30

Back
[00:00:35] -!- guys has quit [Ping timeout: 240 seconds]
[00:27:44] -!- CalimeroTeknik has quit [Ping timeout: 255 seconds]
[00:33:22] -!- CalimeroTeknik has joined #archlinux-ports
[00:46:20] -!- guys has joined #archlinux-ports
[02:17:21] -!- p71 has quit [Ping timeout: 248 seconds]
[02:32:51] -!- p71 has joined #archlinux-ports
[03:43:07] -!- isacdaavid has joined #archlinux-ports
[05:14:36] -!- tyler has joined #archlinux-ports
[05:14:47] tyler is now known as tyzoid_
[05:14:57] tyzoid is now known as tyz
[05:15:05] tyzoid_ is now known as tyzoid
[05:19:01] -!- tyzoid has quit [Client Quit]
[05:53:10] -!- hringriin_ has joined #archlinux-ports
[05:56:41] -!- hringriin has quit [Ping timeout: 276 seconds]
[05:56:41] hringriin_ is now known as hringriin
[06:21:06] -!- deep42thought has joined #archlinux-ports
[06:44:49] -!- p71 has quit [Read error: Connection reset by peer]
[07:39:50] -!- p71 has joined #archlinux-ports
[08:06:11] -!- because has quit [Ping timeout: 268 seconds]
[08:06:50] -!- because has joined #archlinux-ports
[08:23:32] -!- abaumann has joined #archlinux-ports
[08:31:05] <deep42thought> Hi abaumann
[08:31:10] <abaumann> Hi.
[08:31:19] <abaumann> Yes. Let's chat here and not in the forum :-)
[08:31:46] <deep42thought> :-D
[08:31:55] <deep42thought> my latency is a lot smaller on irc
[08:32:42] <abaumann> curl is core, git is extra. both just tested.
[08:32:42] <deep42thought> a patch to automatically mark community packages as tested would be simple
[08:33:14] <deep42thought> y pacman-syu'ed my arch32 box and it auto-tested some of the blocking packages
[08:33:48] <abaumann> git-review is not that critical, and should belong into community IMHO. It's not essential..
[08:33:52] <deep42thought> The list is now https://ptpb.pw
[08:33:54] <abaumann> the rest is community packages.
[08:34:13] <abaumann> libwps..
[08:34:41] <abaumann> Microsoft Works. Give me a break.. :-)
[08:34:54] <abaumann> I would go for the bold move..
[08:34:58] <deep42thought> ok
[08:35:10] <abaumann> me bad if things break. :-)
[08:35:19] <deep42thought> nah, np
[08:35:27] <deep42thought> I've broken things many times this way ;-)
[08:41:35] <deep42thought> btw: here is a sketch of the new database layout: https://buildmaster.archlinux32.org
[08:41:43] <deep42thought> In case I'm doing something stupid :-)
[08:43:37] <abaumann> You know about https://oxygene.sk
[08:43:38] <phrik> Title: Database Modeller - Lukáš Lalinský (at oxygene.sk)
[08:43:51] <deep42thought> no
[08:43:55] <deep42thought> so - now, I do
[08:44:01] <abaumann> :-)
[08:44:11] <abaumann> though I like graphviz...
[08:46:02] <abaumann> yeah. looks good. But I'm not that deep into the package system (yet), so I cannot fully judge.
[08:47:15] <deep42thought> the tricky part is the "install_targets" one (and this is the reason I'm doing this at all): dependencies can be on groups, virtual packages ("provide=...") or real packages
[08:47:22] <abaumann> libreoffice-still 5.3.7-6.1 arived on my stable machine.. all well.
[08:47:31] <deep42thought> :-)
[08:49:15] <abaumann> I was thinking about a manual dependency_type for (yet) unknown dependencies, which can then at least be done manually
[08:50:02] <deep42thought> where would those dependencies be defined/calculated?
[08:50:15] <deep42thought> s/where/how/
[08:51:40] <abaumann> that's exactly the point: imagine a package has a missing depends=() and you add it temporarily and manually to avoid problems.
[08:52:00] <abaumann> on the other hand, the inverse problem also exists..
[08:52:08] <abaumann> maybe that's over-engineering from my side. :-)
[08:52:25] <deep42thought> can't you do both in the modification repository?
[08:52:34] <deep42thought> you need to rebuild the package anyway
[08:52:47] <deep42thought> because otherwise, the depends will not be in the binary package
[08:53:06] <abaumann> true
[08:56:19] <deep42thought> hmm, but maybe we want to lift a "depends" to a "the binary package depends, but the build master should not see the dependency" sometime ...
[08:56:34] <deep42thought> s/lift/relax/
[09:08:52] -!- deep42thought has quit [Quit: Leaving.]
[09:13:11] -!- abaumann has quit [Quit: ChatZilla 0.9.92 [SeaMonkey 2.49.1/20171107234153]]
[10:04:44] -!- isacdaavid has quit [Quit: Leaving.]
[10:12:46] -!- deep42thought has joined #archlinux-ports
[10:42:33] -!- jemadux has joined #archlinux-ports
[11:09:30] -!- brtln1 has joined #archlinux-ports
[11:09:53] -!- wyre has quit [Ping timeout: 255 seconds]
[11:09:58] -!- brtln has quit [*.net *.split]
[11:11:18] -!- wyre has joined #archlinux-ports
[11:12:30] -!- e1z0 has joined #archlinux-ports
[11:12:41] <e1z0> hi
[11:13:37] <deep42thought> Hi
[11:30:24] <e1z0> i've just moved to arch32 and tried to update it, all seems to be worked fine, downloading packages but now it shows that arcives ar corrupted (invalid or corrupted packages (PHP signature)), any ideas?
[11:31:02] <deep42thought> pacman -Sc
[11:31:27] <e1z0> oh just found that it needs archlinux32-keyring-transition package
[11:31:36] <e1z0> btw, thx
[11:32:00] <deep42thought> yeah, that, too
[11:53:50] -!- jemadux_ has joined #archlinux-ports
[12:00:04] -!- jemadux has quit [Remote host closed the connection]
[12:00:04] -!- jemadux_ has quit [Read error: Connection reset by peer]
[12:06:37] -!- FSMaxB has quit [Ping timeout: 255 seconds]
[12:13:47] -!- eschwartz[m] has quit [Ping timeout: 252 seconds]
[12:14:22] -!- z3ntu has quit [Ping timeout: 255 seconds]
[12:14:24] -!- cypa[m] has quit [Ping timeout: 248 seconds]
[12:14:46] -!- brtln1 has quit [Ping timeout: 264 seconds]
[13:27:15] -!- z3ntu has joined #archlinux-ports
[13:44:12] -!- brtln has joined #archlinux-ports
[13:44:12] -!- eschwartz[m] has joined #archlinux-ports
[13:44:13] -!- cypa[m] has joined #archlinux-ports
[13:44:13] -!- FSMaxB has joined #archlinux-ports
[14:02:45] tyz is now known as tyzoid
[14:22:58] -!- guys has quit [Ping timeout: 268 seconds]
[14:33:27] <tyzoid> deep42thought: Not sure what data is relating with what in your diagram
[14:33:34] <tyzoid> also "PHP signature" made me lol
[14:33:44] <deep42thought> me, too
[14:34:44] <deep42thought> tyzoid: yeah, I'm a noob to databases - the arrow indicate, that the index of the tip appears in a column on the other side
[14:35:31] <deep42thought> the information in the bubbles is far from complete
[14:38:21] <tyzoid> let me send you what I had in mind
[14:38:30] <tyzoid> currently putting it in a diagram
[14:41:52] <deep42thought> here is a complete dump of all the columns so far https://buildmaster.archlinux32.org
[14:50:43] <tyzoid> deep42thought: How are you planning on tracking dependencies of built packages?
[14:51:13] <deep42thought> I would not make a difference between a built and a not-yet-built-but-scheduled package
[14:51:23] <deep42thought> scenario is as follows:
[14:52:21] <deep42thought> change in git introduces new version -> new entry in package_sources and binary_packages is generated (from makepkg --printsrcinfo) and new dependencies on already existent packages/groups/virtual_packages are created (in both directions)
[14:52:34] <deep42thought> (probably not in both directions - I have to think about that)
[14:53:15] <tyzoid> https://imgur.com
[14:53:16] <phrik> Title: Imgur: The most awesome images on the Internet (at imgur.com)
[14:53:17] <deep42thought> and whenever I want to retrieve dependencies, I just need to filter through "dependencies" (or "view_dependencies", which is the combined version
[14:53:23] <tyzoid> That's the start of what I was thinking of
[14:53:57] <tyzoid> except dependency_type is erroneously shown as a foreign reference to package.id
[14:54:19] <deep42thought> yeah
[14:54:29] <deep42thought> so the lower part is identical to what I have
[14:54:31] <deep42thought> modulo names
[14:54:34] <tyzoid> the problem is that dependencies can change over time
[14:54:47] <tyzoid> so I'd want to encode the dependency tree for revisions too
[14:54:50] <deep42thought> yes, this is, why I want to track them for binary packages
[14:54:53] <deep42thought> not package sources
[14:55:29] <tyzoid> I could make a revision_dependency table, but that'd grow hilariously large
[14:55:59] <deep42thought> but otherwise you won't catch time dependent dependencies
[14:56:00] <tyzoid> If keyed correctly, it should perform decently, but it'd require a recursive query to dig up the whole tree
[14:56:07] <tyzoid> same with DependsOn
[14:56:46] <tyzoid> because a minimal encoding would defer to dependent packages' dependencies
[14:57:25] <deep42thought> I inserted the "install_targets" table - because you can depend on a group, package or an entry from 'provides='
[14:58:09] -!- abaumann has joined #archlinux-ports
[14:58:15] <deep42thought> Hi Andreas
[14:58:18] <abaumann> Hi :-)
[14:58:41] <tyzoid> wait, you can depend on a group?!?
[14:58:46] <deep42thought> sure
[14:58:48] <deep42thought> base?
[14:58:50] <deep42thought> base-devel?
[14:58:56] * tyzoid facepalms
[15:00:07] <tyzoid> but then we need to store group changes
[15:00:26] <deep42thought> binary packages know to which groups they belong
[15:00:33] <deep42thought> "group_members" in my diagramm
[15:00:51] <tyzoid> deep42thought: I'd rather transform a hard dependency on a group to enumerating each package in that group.
[15:01:25] <tyzoid> deep42thought: But if we're using the database to track this, then we don't want to rely on having the binary package?
[15:01:31] <deep42thought> but then, you won't update that list if some package decides to now belong to that group, too
[15:01:49] <tyzoid> deep42thought: Which is intended for a package revision
[15:01:57] <deep42thought> huh?
[15:02:14] <tyzoid> if package (a) depends on group (y), which did not include package (x) at build time, it does not depend on package (x) at a later time
[15:02:16] <deep42thought> the (needed) information of the binary package should be completely in the database
[15:02:38] <tyzoid> even if package (x) was later included in (y)
[15:02:47] <deep42thought> hmm, ok
[15:03:04] <deep42thought> actually: no
[15:03:05] <tyzoid> whether pacman will accept that argument is another story :P
[15:03:38] <deep42thought> think of: package (b) from (y) gets splitted into (b) and (c) which is now also in (y)
[15:03:52] <deep42thought> then you will still depend on (c)
[15:04:20] <tyzoid> sure, but you may have broken binary compatability at that point: the package would need to be rebuilt anyway
[15:04:44] <deep42thought> it's not only about binary dependencies
[15:04:47] <deep42thought> unfortunately
[15:04:53] <tyzoid> The revisional dependency tracking would also include the version of the dependencies at build time
[15:05:10] <tyzoid> unless we only ever want to encode current information into the database.
[15:05:16] <tyzoid> and not keep historical data.
[15:05:24] <deep42thought> I would purge historical data
[15:05:27] <tyzoid> ok
[15:05:41] <tyzoid> so in that case, storing revision dependency version information is moot
[15:06:08] <deep42thought> well, we might have to track the same package in different revisions
[15:06:18] <tyzoid> so in that case, the only 'revisions' that will be in the database will be the ones in the repo?
[15:06:28] <deep42thought> and on the build list
[15:06:33] <tyzoid> right.
[15:06:34] <deep42thought> (which is treated like a repo)
[15:06:47] <deep42thought> sry, I'll have to leave now
[15:06:49] <tyzoid> so once the package gets replaced with a newer version, we discard the database info
[15:06:49] <tyzoid> okay
[15:06:52] <tyzoid> cya
[15:06:59] <deep42thought> feel free to fill your graph, I'm not really happy with mine either :-)
[15:07:14] -!- deep42thought has quit [Quit: Leaving.]
[15:40:10] -!- znc_tzd_test has joined #archlinux-ports
[15:49:30] -!- znc_tzd_test_ has joined #archlinux-ports
[15:49:31] -!- znc_tzd_test has parted #archlinux-ports
[15:50:59] -!- znc_tzd_test_ has parted #archlinux-ports
[15:58:49] -!- guys has joined #archlinux-ports
[16:05:54] -!- abaumann has quit [Quit: ChatZilla 0.9.92 [SeaMonkey 2.49.1/20171107234153]]
[16:08:10] -!- zoidby has joined #archlinux-ports
[16:10:38] <zoidby> Hi, is it possible, that gnucash needs to be rebuilt? http://ix.io
[16:32:02] <tyzoid> possible? yes.
[16:32:38] <tyzoid> zoidby: Feel free to file a bug for it: https://bugs.archlinux32.org
[16:32:39] <phrik> Title: Flyspray::Archlinux32: Tasklist (at bugs.archlinux32.org)
[16:33:00] <guys> what gnucash package, arch 64 doesn't have it
[16:33:30] <tyzoid> guys: https://pkgapi.arch32.tyzoid.com
[16:35:58] <guys> tyzoid: that package exists in svntogit, but it's only in trunk
[16:38:24] <tyzoid> zoidby: https://aur.archlinux.org
[16:38:24] <phrik> Title: AUR (en) - gnucash (at aur.archlinux.org)
[16:38:41] <tyzoid> Looks like new versions of gnucash require installing icu 58
[16:38:44] <tyzoid> 59*
[16:39:14] <zoidby> tyzoid: Oh thanks, i’ll try
[16:39:19] <guys> tyzoid: why do you also have webkitgtk2
[16:39:59] <tyzoid> zoidby: https://aur.archlinux.org
[16:39:59] <phrik> Title: AUR (en) - icu59 (at aur.archlinux.org)
[16:40:11] <tyzoid> guys: Because it's a dependency?
[16:42:34] <guys> But... it was dropped from the 64-bit repos and completely purged from svn and svntogit, and you're not supplying your own in https://github.com so where does it come from?
[16:42:35] <phrik> Title: GitHub - archlinux32/packages: package customizations and pure-i686 packages (at github.com)
[16:43:02] <tyzoid> guys: probably was built and never purged.
[16:43:13] <tyzoid> We don't currently purge packages, afaik
[16:43:16] <guys> Ah.
[16:43:43] <guys> So you purge them manually when people report it doesn't work, and you're unable to rebuild it because the dependencies are gone?
[16:43:55] <tyzoid> pretty much.
[16:44:19] <tyzoid> Once automation catches up a bit, we'll be able to do more.
[16:44:37] <tyzoid> but it's slow, seeing as deep42thought pretty much maintains it on his own.
[16:44:46] <tyzoid> he's quite busy, as you can imagine
[16:45:13] <tyzoid> guys: Do you know why it was removed?
[16:45:21] <tyzoid> was it because it still required webkitgtk2?
[16:46:14] <guys> tyzoid: yeah
[16:46:27] <guys> And we purged webkitgtk2 and all reverse depends
[16:55:37] <zoidby> tyzoid: Installing icu59 from the aur did the trick. But installing aur dependencies for official packages shouldn't be necessary. Wouldn't it be possible, to include icu59 in the dependencies of the official gnucash package and add icu59 to the official repos?
[16:56:09] <tyzoid> zoidby: The problem is that arch upstream removed them from their repos, so we no longer build them.
[16:56:23] <tyzoid> The fact that they're in our repos is just because we haven't cleaned them up yet.
[16:56:27] <zoidby> gnucash?
[16:56:39] <tyzoid> yeah, gnucash is now only available in the AUR.
[16:56:52] <zoidby> pfff, ok
[16:56:55] <tyzoid> zoidby: See guys' comment above.
[16:56:59] <tyzoid> as to why
[17:11:34] -!- zoidby has quit [Quit: leaving]
[17:21:00] -!- krumelmonster has joined #archlinux-ports
[17:33:42] -!- abaumann has joined #archlinux-ports
[17:35:57] -!- abaumann has parted #archlinux-ports
[17:39:30] -!- p71 has quit [Read error: Connection reset by peer]
[17:43:45] -!- p71 has joined #archlinux-ports
[18:40:07] -!- abaumann has joined #archlinux-ports
[18:44:36] -!- isacdaavid has joined #archlinux-ports
[18:45:47] <abaumann> plugins
[18:46:47] <abaumann> darn... new to irssi :-)
[19:27:55] -!- abaumann has quit [Quit: leaving]
[19:49:41] -!- krumelmonster has quit [Remote host closed the connection]
[19:51:49] -!- Polichronucci has quit [Read error: Connection reset by peer]
[19:57:19] -!- abaumann has joined #archlinux-ports
[20:00:43] -!- abaumann has quit [Client Quit]
[20:21:27] -!- abaumann has joined #archlinux-ports
[20:50:19] -!- Polichronucci has joined #archlinux-ports
[21:40:50] -!- drathir has quit [Ping timeout: 252 seconds]
[21:49:06] -!- abaumann has quit [Quit: leaving]
[22:16:26] -!- deep42thought has joined #archlinux-ports
[22:25:38] <deep42thought> ok, I removed gnucash and gnucash-docs
[22:25:49] <deep42thought> interestingly, the build master did _not_ put it on the deletion-list
[22:26:01] <deep42thought> (but that list does not get evaluated yet anyway)
[22:28:45] <deep42thought> guys: regarding not-yet-implemented deletion: I thought, it is not that urgent to properly delete old packages - it's more important to not delete recent packages :-)
[22:29:19] <deep42thought> It happened already 3 or 4 times, that all packages entered the deletion-list
[22:29:31] <deep42thought> imagine the fun if automatic deletion was already set up ...
[22:53:49] -!- isacdaavid has quit [Quit: Leaving.]
[23:00:42] <deep42thought> ah, I stand corrected: you can _not_ depend on a group: makepkg will install the group and then error with '==> Missing dependencies: -> qt5'
[23:36:44] -!- deep42thought has quit [Quit: Leaving.]