#archlinux-ports | Logs for 2017-11-29

[00:08:14] <rewbycraft> Yeah/exit
[00:08:17] <rewbycraft> Dangit
[00:31:54] -!- xes_ has joined #archlinux-ports
[00:32:57] -!- xes has quit [Ping timeout: 240 seconds]
[01:23:41] -!- guys has quit [Ping timeout: 240 seconds]
[01:51:01] -!- guys has joined #archlinux-ports
[02:46:40] -!- isacdaavid has quit [Quit: Leaving.]
[04:58:02] -!- yokel has quit [Remote host closed the connection]
[05:00:39] -!- yokel has joined #archlinux-ports
[05:53:47] -!- hringriin_ has joined #archlinux-ports
[05:56:52] -!- hringriin has quit [Ping timeout: 260 seconds]
[05:56:52] hringriin_ is now known as hringriin
[06:39:11] -!- deep42thought has joined #archlinux-ports
[07:34:55] -!- deep42thought has quit [Quit: Leaving.]
[08:16:52] -!- abaumann has joined #archlinux-ports
[08:48:18] -!- abaumann has quit [Quit: Page closed]
[09:00:40] -!- deep42thought has joined #archlinux-ports
[09:38:25] xes_ is now known as xes
[14:13:03] -!- guys has quit [Ping timeout: 248 seconds]
[14:56:31] -!- guys has joined #archlinux-ports
[16:38:36] -!- deep42thought has quit [Quit: Leaving.]
[16:39:43] -!- yokel has quit [Ping timeout: 248 seconds]
[16:40:59] -!- yokel has joined #archlinux-ports
[17:42:32] -!- deep42thought has joined #archlinux-ports
[17:44:33] <deep42thought> tyzoid: FYI: I'm considering to move the build master's internal state directory to a mysql database - then you can connect your package info api to that.
[17:45:13] <tyzoid> deep42thought: Sounds like a good idea.
[17:45:36] <tyzoid> deep42thought: Perhaps we could also set up sql replication? Cross-continent mysql connections tend to be a bit slow
[17:46:08] <deep42thought> I have 0 experience with that
[17:46:21] <deep42thought> but you could also put the package api onto the buildmaster
[17:46:39] <tyzoid> deep42thought: How does the complexity of that compare to having a REST api on the buildmaster itself
[17:46:47] <tyzoid> and by that I mean generating static json files?
[17:47:27] <tyzoid> If you're doing it for other reasons, then it's nice. I would not recommend doing it just for my API.
[17:47:55] <deep42thought> no, I'm doing it to improve some ugly stuff on the build master
[17:48:10] <tyzoid> Ok, are you also considering moving to something like python?
[17:48:18] <deep42thought> this could be step 2
[17:48:30] <tyzoid> Well mysql + bash is a disaster waiting to happen.
[17:48:30] <deep42thought> but first I want to put the state into a database
[17:48:45] <deep42thought> not worse than bash + a directory full of files
[17:49:11] <tyzoid> actually, yes, because you have to rely on parsing the mysql-client output
[17:49:19] <tyzoid> bash + files is better than bash + sql
[17:49:55] <deep42thought> bash+files can be pretty ugly, too - if your "querys" start to become complex
[17:50:06] <tyzoid> true
[17:50:14] <tyzoid> but still
[17:50:39] <tyzoid> up to you, though.
[17:50:52] <deep42thought> bash + sql will be only temporary
[17:50:56] <tyzoid> deep42thought: If you want to start small, just write the python wrapper for the mysql db
[17:51:03] <deep42thought> have not decided yet, what the 'front end' will become
[17:51:06] <tyzoid> then wrap python in bash functions to grab the data
[17:51:23] <tyzoid> because then you can generate a stable api that's easy to use in bash
[17:51:34] <deep42thought> my plan is actually different:
[17:51:48] <deep42thought> set up the database as "slave", e.g. generated from the state directory
[17:52:04] <deep42thought> then transfer all the "read" logic to mysql
[17:52:10] <deep42thought> and then all the write logic
[17:52:24] <tyzoid> sounds like that should work
[17:52:33] <deep42thought> and the steps in between will be most likely bash called python/whatever scripts
[17:52:35] <tyzoid> but I still recommend writing a python wrapper for all sql functions.
[17:52:59] <tyzoid> I would recommend not doing bash + sql at all
[17:53:17] <deep42thought> what's the advantage of havin python in between?
[17:53:31] <tyzoid> That you don't need to write sql in bash
[17:53:38] <tyzoid> or parse mysql-client's output
[17:53:48] <deep42thought> printf '' | mysql
[17:53:49] <deep42thought> ?
[17:53:51] <deep42thought> bad idea?
[17:54:06] <tyzoid> yes, because you have the opportunity for SQL injection
[17:54:18] <tyzoid> use prepared statements in pyton.
[17:54:25] <deep42thought> right
[17:54:34] <tyzoid> s/pyton/python/
[18:34:35] <r00t^2> tyzoid: everything sorted with your mirror? i remember seeing some issues with it the other day when my local mirror was down
[18:35:05] <tyzoid> r00t^2 Should be. My domain registrar had a ~8min DNS outage.
[18:35:22] <r00t^2> oh criminy :/ sorry to hear. glad it's all sorted though!
[18:35:24] <tyzoid> Sunday morning at ~11AM ET.
[18:35:27] <r00t^2> wow.
[18:35:37] <r00t^2> which registrar is that?
[18:35:43] <tyzoid> Namecheap.
[18:35:51] <tyzoid> They're very good in general, though.
[18:35:58] <r00t^2> oh yeah,
[18:36:06] <tyzoid> First DNS outage I've experienced in my 4 years being with them.
[18:36:12] <r00t^2> i don't use them for nameservers, but i do most of my registrations through them
[18:36:19] <tyzoid> first outage of any kind, I should say.
[18:36:23] <r00t^2> i just wish that had more gTLDs
[18:36:28] <r00t^2> they they, rather
[18:37:21] <r00t^2> and i'm bummed they don't offer thawte certs anymore, but less of an issue now that Let's Encrypt has gotten most of their kinks out
[18:47:52] <tyzoid> r00t^2: Have you seen the nginx config which automatically renews the letsencrypt cert?
[18:48:33] <r00t^2> i haven't seen the config that does that, i think, but i cooked something up that handles it with the manual --certonly stuff
[18:49:56] <tyzoid> I currently use https://github.com
[18:49:57] <phrik> Title: GitHub - diafygi/letsencrypt-nosudo: Free HTTPS certificates without having to trust the letsencrypt cli with sudo/root (at github.com)
[18:50:06] <tyzoid> which allows me to renew certs without root.
[18:52:45] <r00t^2> iiinteresting, i'll have to look through that and see how they do it
[18:54:19] <r00t^2> "There is only one command that needs to be run as root on your server and it is a very simple python https server that you can inspect for yourself before you run it." you know, i wonder if you could even get around what with CAP_NET_BIND_SERVICE
[18:54:36] <r00t^2> s/what with/that with/
[18:55:29] <r00t^2> oh, setcap works on binaries, not scripts... you'd probably have to bytecompile the program and then grant the cap on that
[18:55:30] <tyzoid> r00t^2 I got around it by changing my port 80 to port 8082
[18:55:37] <r00t^2> lol, that works too
[18:55:38] <tyzoid> so it binds to a different port.
[18:55:55] <tyzoid> Apache listens on 80 anyway, so I just ProxyPass all requests to that path to 8082
[18:56:05] <r00t^2> or you could do a redirection
[18:56:09] <r00t^2> oooooh, clever!
[18:56:46] <tyzoid> r00t^2 I've got a set of apache management scripts on my server.
[18:56:54] <tyzoid> mksite, mkproxy, encsite, etc.
[18:57:25] <r00t^2> anything in particular that draws you to apache over nginx?
[18:57:49] <tyzoid> other than the fact that I started using apache back in 2012?
[18:57:54] <tyzoid> I first learned on apache
[18:57:58] <tyzoid> so all my scripts are apache-focused
[18:58:39] <tyzoid> that said, I use nginx if I need a system to just do reverse-proxying
[18:59:04] <tyzoid> it's a bit lighter weight
[18:59:38] <tyzoid> r00t^2: Another thing, is apache uses mod_php built in, so I don't need to manage issues between php-fpm and nginx
[18:59:42] <tyzoid> since it's all integrated
[18:59:53] <tyzoid> which means all my fastcgi() function calls actually work.
[19:01:17] <r00t^2> heh, that does simplify things
[19:40:50] -!- abaumann has joined #archlinux-ports
[19:46:01] <deep42thought> abaumann: I'm planning on moving the build master's backend to mysql - then I would also introduce an "arch" variable for the packages
[19:46:18] <abaumann> ah. ok.
[19:46:19] <deep42thought> and then we can start building and distributing i486
[19:46:35] <abaumann> wow. hang on a mo. I'm not even sure it will work at all.
[19:46:39] <abaumann> ok. :-)
[19:46:43] <deep42thought> abaumann: np
[19:46:49] <abaumann> maybe a i585 then.
[19:46:59] <deep42thought> would be enough for _me_
[19:47:09] <abaumann> I'm still strugling to get the gettext, pkg-build, glibc, meson disaster done.
[19:47:34] <abaumann> mine is a K6 without CMOV, so it would also be ok for me.
[19:48:12] <deep42thought> https://ptpb.pw
[19:49:53] <abaumann> .it has cmov. :-)
[19:53:55] <abaumann> this sounds like an old IGEL or something alike...
[19:54:04] <deep42thought> alix
[19:55:00] <abaumann> CPU: 500 MHz AMD Geode LX800
[19:55:13] <abaumann> oh, right. I never bothered to check on my router. But yeah. :-)
[19:55:21] <deep42thought> :-D
[21:22:19] <abaumann> sweet: gettext has a depends=(glib2) and it's not even mandatory..
[21:39:05] <guys> file a bugreport
[21:39:35] <abaumann> i will. but let me first see, maybe some parts of gettext need parts of glib2 for something..
[21:40:06] <abaumann> ..and may build breaks further down in step 11 or so. :-)
[21:40:15] <abaumann> s/may/my
[21:41:48] <guys> /usr/lib/libgettextsrc- uses libglib
[21:42:22] <guys> So does /usr/lib/libgettextlib-
[21:42:41] <abaumann> yeah. but what functionallity inside the code needs glib2 to work
[21:43:02] <guys> So everything does contain link references, but as for why I couldn't say, since I don't know why it shouldn't either
[21:44:13] <guys> https://git.savannah.gnu.org
[21:44:14] <phrik> Title: DEPENDENCIES - gettext.git - GNU gettext (at git.savannah.gnu.org)
[21:44:23] <abaumann> ah. configure says: --with-included-glib
[21:44:34] <abaumann> so fair enough.. normally you would link against the system glib
[21:44:47] <abaumann> and in case of bootstrapping you have to read the manual :-)
[21:45:03] <abaumann> (y)
[21:45:05] <abaumann> thanks.
[21:45:20] <abaumann> If not present, a subset of libglib2 (included in this package) will be
[21:45:21] <abaumann> compiled into libgettextlib.
[21:45:54] <abaumann> actually nice. glib2 is anyway a too big library to link to anything most likely (but gnome).
[21:46:07] <abaumann> it should have been split into 20 libraries or so years ago.
[21:46:42] <abaumann> wc -l configure.ac of glib2 is 3523 lines. this says it all :-)
[21:49:51] <guys> I was given to understand that glib2 was the type of library that should be shot at birth, and its users shunned. Who cares about *splitting* it o_O
[21:50:08] <abaumann> :)
[22:04:14] -!- xes_ has joined #archlinux-ports
[22:06:54] -!- xes has quit [Ping timeout: 255 seconds]
[22:10:01] -!- abaumann has quit [Remote host closed the connection]
[23:50:05] -!- deep42thought has quit [Quit: Leaving.]