#archlinux32 | Logs for 2018-04-14

[00:04:05] -!- davor has joined #archlinux32
[01:15:21] -!- eduardoeae_ has joined #archlinux32
[04:47:10] -!- Mart|n has quit [Remote host closed the connection]
[07:39:05] -!- oaken-source has joined #archlinux32
[07:56:05] -!- davor has quit [Ping timeout: 240 seconds]
[08:01:20] -!- davor has joined #archlinux32
[08:10:29] -!- eduardoeae has quit [Ping timeout: 260 seconds]
[08:28:17] -!- esotericnonsens_ has joined #archlinux32
[08:28:28] -!- esotericnonsense has quit [Read error: Connection reset by peer]
[09:28:54] -!- belanthor has joined #archlinux32
[09:35:09] -!- davor has quit [Ping timeout: 264 seconds]
[09:40:16] -!- davor has joined #archlinux32
[10:54:26] -!- abaumann has joined #archlinux32
[10:58:28] -!- deep42thought has joined #archlinux32
[11:01:31] <abaumann> deep42thought: hi :-)
[11:01:37] <abaumann> https://bbs.archlinux32.org
[11:01:38] <phrik> Title:[SOLVED]Chromium not working after system update / Pacman / Pacman Upgrades / Arch Linux 32 Forums (at bbs.archlinux32.org)
[11:01:39] <deep42thought> Hi abaumann
[11:01:44] <abaumann> I just tries to stabilize qt5-base
[11:01:53] <abaumann> Seems we still have some icu.60 issues.
[11:02:34] <abaumann> *tried
[11:02:51] <deep42thought> https://packages.archlinux32.org
[11:02:52] <phrik> Title:More and less critical issues with the database (at packages.archlinux32.org)
[11:02:53] <deep42thought> says so, too
[11:04:14] <abaumann> mmh textlife-bin was one package I triggered manually, and it still has a icu.60? *puzzle*
[11:04:46] <deep42thought> the information on that site is from the objdump-run on the package
[11:04:57] <deep42thought> so maybe it's in a file, which is not loaded actually?
[11:05:11] <deep42thought> (we also see links to x86_64 libraries, sometimes)
[11:06:51] <deep42thought> where do you see textlive-bin on that site??
[11:07:14] <abaumann> extra/texlive-bin-2017.44590-10.1-i686 depends on libicuuc.so.60 which is not provided by any package - but can be replaced by the one in testing.
[11:07:18] <abaumann> extra/texlive-bin-2017.44590-10.1-i686 depends on libicui18n.so.60 which is not provided by any package - but can be replaced by the one in testing.
[11:07:34] * deep42thought cannot type properly
[11:07:40] <abaumann> :-)
[11:07:53] <deep42thought> I was looking for "textlive" ...
[11:08:28] <abaumann> yeah. beacause I also mispelled it :-)
[11:08:42] <abaumann> 11:04 < abaumann> mmh textlife-bin was one package I triggered manually, and it still has a icu.60?
[11:08:45] <abaumann> *puzzle*
[11:08:48] <deep42thought> oh, I see
[11:08:56] <deep42thought> it's in testing
[11:09:04] <deep42thought> and that one is not listed, so: has no issues
[11:09:06] <abaumann> yep.
[11:09:13] <abaumann> same for qt5-base.
[11:09:19] <abaumann> they work. I tested.
[11:09:28] <deep42thought> I'll run "db-update --progressive"
[11:09:38] <deep42thought> let's see if it fixes more than it breaks ...
[11:09:49] <abaumann> ok. tex has a weird issue wie a SSE2 optimization in some Lua code, but I think it's not fatal.
[11:09:58] <abaumann> :-)
[11:27:34] <deep42thought> hmm, the db-update command still runs :-/
[11:28:21] <deep42thought> but I really fear, that my stored function used by db-update is a) not 100% correct and b) for sure not optimal
[11:28:27] <deep42thought> someone should revise it ...
[11:28:41] <deep42thought> ... someone with more knowledge of databases than me
[11:41:32] <deep42thought> ok, the command finished
[11:41:42] <deep42thought> there are still some libicu-60s linked ...
[11:41:44] <abaumann> qt5-base arived in stable :-)
[11:56:28] -!- eduardoeae has joined #archlinux32
[12:27:55] -!- abaumann has quit [Quit: leaving]
[12:29:45] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.stdin.2018-04-14T12:28:05.jOvuB5"?.
[12:52:48] * buildmaster resumes sanity.
[13:37:37] <deep42thought> abaumann, tyzoid, everyone: I think, the package frontend is now ready for testers: packages.archlinux32.org - feel free to tell me about minor and major issues/wishes - or open a pull request on github.com/archlinux32/archweb32 .
[14:03:45] -!- davor has quit [Ping timeout: 255 seconds]
[14:10:22] -!- davor has joined #archlinux32
[14:23:50] -!- abaumann has joined #archlinux32
[14:24:07] <abaumann> deep42thought: looks good. :-)
[14:24:17] * deep42thought is proud
[14:24:27] <deep42thought> but I failed the .css test
[14:24:41] <abaumann> mmh?
[14:24:51] <deep42thought> if I remove the empty "packagefiles" <div> at the end of the page, the blue box is wayyy to short
[14:24:58] <deep42thought> :-/
[14:25:52] <abaumann> ah. I found some broken links: https://packages.archlinux32.org click to i686, core
[14:26:20] <deep42thought> ups
[14:27:23] <deep42thought> should be fixed
[14:27:35] <deep42thought> (but the wrong might be cached a minute or two)
[15:23:49] <tyzoid> deep42thought: https://packages.archlinux32.org
[15:23:50] <phrik> Title:Arch Linux 32 - 0ad a22-6.0 (i686) (at packages.archlinux32.org)
[15:23:53] <tyzoid> icu shows up twice in the dep list
[15:24:23] <tyzoid> as do a few other link deps
[15:24:47] <deep42thought> there are several "legit" reasons, why dependencies show up twice
[15:24:58] <deep42thought> mesa shows up twice, because it's in testing and extra
[15:25:32] <deep42thought> and the each type of dependency does not exclude another type of dependency to the same package
[15:26:09] <deep42thought> and then, "link" dependencies may be duplicate, because multiple libraries are linked
[15:26:25] <deep42thought> I think, the only one of these 3, which we should change is the later one
[15:26:28] <deep42thought> do you agree?
[15:27:19] <deep42thought> hmmm
[15:27:45] <deep42thought> something seems wrong, though: link dependencies should have a "*.so" as install_target (the thing printet before the brackets
[15:27:55] <deep42thought> so "icu (link)" is wrong ...
[15:30:25] <tyzoid> Well we're listing the package as it is in that repo
[15:30:35] <tyzoid> So the package, and it's dependencies should be in that repo
[15:30:37] <tyzoid> ideally
[15:30:37] <tyzoid> right?
[15:30:49] <deep42thought> installable from a system running on that repo
[15:30:59] <deep42thought> e.g. a testing package may have core dependencies
[15:31:02] <tyzoid> right
[15:31:05] <tyzoid> yeah
[15:31:17] <deep42thought> but on the other hand, you can also have a core package on a testing system
[15:31:19] <deep42thought> :-/
[15:31:29] <tyzoid> so we should only show the dependent package that would be installed along with the package from that repo
[15:31:50] <deep42thought> if you come up with a mysql query, I'm more than happy to implement it
[15:31:51] <deep42thought> :-D
[15:31:56] <tyzoid> i.e. if pacman.conf has testing, core, extra, community
[15:32:01] <tyzoid> and you did a pacman -Si on the dep
[15:32:06] <tyzoid> that's the dep/repo that should show up
[15:32:09] <deep42thought> I know what you mean
[15:32:12] <tyzoid> ok
[15:32:14] <tyzoid> :)
[15:32:19] * tyzoid needs to look at the db/queries
[15:32:25] <deep42thought> It's easy to get all candidates from the database
[15:32:27] <tyzoid> btw, I added packages to the header on forum and news
[15:32:34] <deep42thought> but it's hard (for me) to get the one, pacman would chose
[15:32:47] <deep42thought> perfect!
[15:32:57] <tyzoid> Now just to add it to home/download
[15:33:03] <deep42thought> you can also push an updated archlinux.org if you like
[15:33:31] <deep42thought> with packages included, I mean
[17:08:50] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.stdin.2018-04-14T16:58:15.8RTDm5"?.
[17:13:39] -!- isacdaavid has joined #archlinux32
[18:09:10] -!- belanthor has quit [Quit: Leaving]
[18:15:36] -!- abaumann has quit [Quit: leaving]
[19:34:17] -!- davor has quit [Ping timeout: 265 seconds]
[19:40:08] -!- davor has joined #archlinux32
[19:50:01] <tyzoid> deep42thought: I've never seen the base64 wrap technique through sql
[19:50:06] <tyzoid> that's interestin
[19:50:08] <deep42thought> :-/
[19:50:09] <tyzoid> intesresting*
[19:50:28] <deep42thought> as I might have mentioned: I'm a php- and mysql- noob
[19:50:43] <tyzoid> It's not "wrong" or "bad", just different
[19:50:51] <tyzoid> just as secure as the other methods
[19:51:26] <deep42thought> yeah, but noob => non-standard solutions :-D
[19:51:40] <tyzoid> => interesting solutions
[19:51:59] <tyzoid> I would recommend moving to prepared statements, though
[19:52:25] <deep42thought> in php or bash?
[19:52:32] <tyzoid> php
[19:52:44] <tyzoid> $mysql->prepare
[19:53:10] <tyzoid> The general idea is that mixing data and sql can lead to some poor results
[19:53:23] <tyzoid> especially since most people have no idea what the security implications are / how to avoid them
[19:53:35] <tyzoid> so prepared statements allows you to put the data and sql separate
[19:53:41] <deep42thought> and I think, performance will be worse with queries
[19:53:42] <tyzoid> and send them separately to the sql server
[19:54:01] <tyzoid> with non-prepared queries, you mean?
[19:54:05] <deep42thought> yes
[19:54:15] <tyzoid> Performance is about the same, unless you're repeatedly running the same query on the connection
[19:54:20] <tyzoid> since you can re-use the same prepared statement
[19:54:58] <deep42thought> well, base64 adds 25% traffic
[19:55:03] <tyzoid> True, yes
[19:55:15] <tyzoid> but the amount of data is small compared to the (a) returned data, and (b) query
[19:55:31] <tyzoid> maybe adds 5% to the total query, and adds 0% to the returned data
[19:55:32] <deep42thought> and I think, it might make the query slower, when the server actually has to check, that it's data coming in (and not the end of the query or something else)
[19:55:42] <tyzoid> miniscule
[19:55:50] <tyzoid> it'll be overshadowed by disk accesses for the data
[19:56:15] <tyzoid> a base64 decode is insanely fast
[19:59:12] <tyzoid> Have you ever looked at how the query optimizer looks at your query?
[19:59:23] <deep42thought> nope
[19:59:43] <tyzoid> If you open the mysql console, and type in your query, it'll give back the results
[19:59:53] <tyzoid> but if you run the same query, but put `explain` in front of it
[19:59:59] <tyzoid> mysql will tell you how it intends to run the query
[20:00:03] <tyzoid> what indexes it'll use, etc.
[20:00:08] <deep42thought> ah, I tried that once
[20:00:16] <deep42thought> but the output didn't really tell me much :-/
[20:00:20] <tyzoid> really useful for seeing how to make slow queries fast
[20:00:55] <tyzoid> i.e. "There's an index right there, but it's not using it :/"
[20:01:55] <deep42thought> btw: I think, I just got rid of the duplicate dependencies :-)
[20:02:26] <tyzoid> nice
[20:02:52] <deep42thought> now I just need to get rid of duplicate "Required By"s ,too ;-)
[20:03:09] <tyzoid> Uh... https://packages.archlinux32.org https://ptpb.pw
[20:03:33] <deep42thought> yes, it checks, that mysql and json give similar results
[20:03:37] <deep42thought> apparently they don't
[20:04:33] <tyzoid> Though I do find it interesting that buildmaster loads images / stuff from packages.archlinux32
[20:04:39] <deep42thought> maybe my fix was not a fix at all :-/
[20:04:42] <deep42thought> ah, I know!
[20:05:00] <tyzoid> Like "hmm, this is faster on the other server for some reason"
[20:05:02] <deep42thought> I wanted to shovel some load off that machine
[20:05:15] <tyzoid> Yeah, just thought it was funny
[20:05:32] <deep42thought> well, that's information from the database
[20:05:36] <tyzoid> btw, planning on migrating ZNC tonight
[20:05:38] <deep42thought> so technically not "buildmaster"
[20:05:41] <deep42thought> ok
[20:07:03] <tyzoid> No need to do anything extra, but you may need to ghost at some point afterward.
[20:07:13] <tyzoid> The plan is to leave both running, but switch over the DNS record
[20:07:14] <deep42thought> hah!
[20:07:22] <deep42thought> https://packages.archlinux32.org works again
[20:07:24] <phrik> Title:Arch Linux 32 - 0ad a22-5.0 (i686) (at packages.archlinux32.org)
[20:07:53] <tyzoid> nice
[20:08:07] <deep42thought> there's still some error ...
[20:09:22] <deep42thought> hmm, removing duplicates from "Required By"s is a little harder ...
[20:10:51] <tyzoid> why not just deduplicate in php?
[20:11:19] <deep42thought> it may be possible, that multiple versions of package x depend on package y - e.g. if y is in [core] and x-1 in [extra] and x-2 in [testing] (and no other versions of y available).
[20:11:34] <deep42thought> tyzoid: because I don't like pulling more data than necessary from mysql
[20:13:52] <tyzoid> can't you group by package name?
[20:14:08] * tyzoid hasn't had a chance to read through/mess with the queries, thoguh
[20:14:11] <tyzoid> though*
[20:17:01] -!- abaumann has joined #archlinux32
[20:17:14] <deep42thought> Hi Andreas!
[20:17:25] <abaumann> That's a welcome :-)
[20:17:27] <abaumann> hi.
[20:17:41] <deep42thought> I was just thinking of you, when eating my Toblerone ;-)
[20:18:05] <abaumann> ...and I don't even particularly like it..
[20:18:22] <deep42thought> :-D
[20:21:02] <deep42thought> tyzoid: is it in general better to put conditions in mysql joins to "on" or to "where" or does it depend on the specific case?
[20:21:27] <deep42thought> "better" = "faster"
[20:29:47] <deep42thought> \o/ no duplicates in "Required By" anymore :-)
[20:31:28] <deep42thought> well, ther _may_ be duplicates, still, but (I believe) they're real, then.
[20:31:36] <deep42thought> s/ther/there/
[20:46:48] <deep42thought> oh, I list packages on the build-list as "Required By", too
[20:48:50] <deep42thought> well, anyway, time for my bed :-)
[20:49:06] -!- deep42thought has quit [Quit: Leaving.]
[20:57:21] * buildmaster resumes sanity.
[21:59:21] -!- abaumann has quit [Quit: leaving]
[22:28:34] -!- oaken-source has quit [Quit: leaving]
[23:09:35] <tyzoid> deep42thought / girls: Depends on the case, though IIRC, if there is a difference, it'll usually be in the favor of the 'on' clause.
[23:38:58] -!- isacdaavid has quit [Quit: Leaving.]