IRC Logs for #crux-devel Sunday, 2015-01-25

Romsteractually wine 1.7.35 and i started bumping compat-32 stuff just hadn't finished it yet01:40
Romsterhrm so that gentoo mirror is out of sync01:42
Romsterthis mirror setup we have now could introduce a security situration if one of us downloads a tampered file by mistake from a different mirror than upstream project.01:51
Romsterand then one of us uses that md5sum file when we build a new Pkgfile version.01:51
RomsterI always make sure i get my files from source= and not some mirror when building new versions.01:52
Romsterits only a small chance since most of us wont have the new version before we push the updated Pkgfile.01:53
*** _mavrick61 has joined #crux-devel02:09
*** diverse1 has joined #crux-devel02:10
*** diverse1 has quit IRC02:10
*** mavrick61 has quit IRC02:16
*** diverse has quit IRC02:16
*** novak has quit IRC02:16
*** novak has joined #crux-devel02:16
RomsterteK_, add dri3proto-1.0.tar.bz2 to that list of messed up ness..02:25
*** tvaalen has joined #crux-devel02:25
Romsteri'm just gonna not use the mirror pool on until that is fixed.02:25
*** frinnst_ has joined #crux-devel02:27
*** diverse has joined #crux-devel02:29
*** frinnst has quit IRC02:35
*** tvaalen_ has quit IRC02:35
*** pitillo_ has joined #crux-devel02:45
*** pitillo has quit IRC02:49
*** _mavrick61 has joined #crux-devel03:41
*** _mavrick61 has quit IRC03:48
*** _mavrick61 has joined #crux-devel03:49
*** hnode_ has joined #crux-devel04:13
*** hnode_ has quit IRC04:48
diversecool, check the diff and see if you are happy06:05
diversemeanwhile there were users complaining about not being about to go to via elinks and there was a problem getting to the wiki with lynx because it just says "we are under maintenance blah"06:07
Romsteri already read the backlot06:18
Romsterthe site isn't my concern since i can't do anything with it.06:29
Romsterbut i do maintain lynx06:29
*** Feksclaus has joined #crux-devel06:50
diverseRomster: the help command does work, I tested it07:39
diverseso users have 2 ways to referring to help, `mport help [command]` and `mport <command> --help`07:40
diversedepending what suits them or what is more convenient07:40
Romsterah that's good07:43
Romsteri was thinking you could work on the lock and unlock command.07:43
Romsterthat wont take much effort07:43
diverselock and unlock?07:43
Romsteryeah prt-get lock unlock and listlocked07:44
diverseI guess I haven't looked into that feature before07:44
Romster  lock <port1 port2...>             lock current version of packages07:44
Romster  unlock <port1 port2...>           unlock packages07:44
Romster  listlocked                        list locked packages07:44
diverseah, okay07:44
Romsteron prt-get07:44
Romsterjsut saves package name to a file in /var/lib/pkg/07:45
Romsterand skips that on sysup/update etc.07:45
diversewhich file is that?07:45
Romsteroccasionally you want to hold back to a earlier version07:45
diverseI got that07:46
Romsterlikei locked them two due to a older video card i have.07:46
Romsterjust use the same file for compatibility for now.07:48
diverseI know07:49
Romstersame format but just names no empty lines07:49
diverse(it would be nice if I could do a file instead though)07:49
Romstercould od that later or maybe you can do a define for prt-get/pkgutils compatabilty mode and a import tool to move to the new files/formats07:50
diverseso does the lock/unlock command check if the port exists or just add whatever the user puts in?07:50
Romsterit adds and removes from that file.07:51
diversedoes the locker file have to be sorted?07:51
Romsterand whats in that file prt-get skipps on sysup and prt-get diff07:51
Romsteri don't believe so07:52
Romsterfirst served07:52
Romsterwouldn't matter you'd read that file before a update/sysup and then skip them ports in the tree.07:52
Romsterso read once store in meory literate though that on every port in a update/sysup/diff and ignore if found.07:53
diverseI guess a HashSet should do the trick07:53
Romsteryeah that would e fast07:53
diverseHashes are faster compared to their Tree/BTree counterparts if ordering doesn't really matter in most cases07:54
Romsterdoes rust have a is in list kinda thing so it an do a btree scan of an array so you don't have to loop through every element?07:55
diverseyeah, the .get() method07:55
Romsterreally any name in that file is ignored on subsequent commands excerpt for them 3 commands07:55
Romsterlistlocked would need to read and display though.07:56
diverseyeah just a reader07:56
Romsterwhich should be a sorted list to display i guess.07:56
diversethat would be a bit slower, but that doesn't really matter07:57
diverseI guess I wouldn't need a reader, just iterate through the Set and print07:57
Romsterits just for display and really that file wont get that big07:58
Romsterin practice usually under 10 entries.07:58
diverseyeah it's not a big deal07:58
Romsterbut it could have more depending on the user.07:58
diverseit's not a big deal, the Set is in memory anyway07:59
Romsterone of the easier things to implement.08:00
Romsteranother thing but do that later is prt-get.aliases08:00
Romsterthat design is a pain in the butt in prt-get how it is.08:00
diversesounds like a pain08:01
Romsterwhat that really needs is a /etc/mport/aliases.d/08:01
Romsterso a port like exim can add a file called /etc/mport/aliases.d/exim containing mail. so any port that # Depends on: mail will say i have exim installed. in the deptree and prt-get info exim or prt-get info mail08:02
Romsterwhat happens now is there is a few aliases fixed in /var/lib/pkg/prt-get.aliases08:03
Romsterand when i did start to check if there or not and add to it in pre-post-install scrips i got told off about doing that as there was no real way to check if that file be consistent.08:03
Romstereven though i was checking if entry existed before adding08:04
Romsterif it was a directory it be easier to get a file added in the $PKG to do it.08:04
diversehmm, maybe what I would like is to have the directory in /etc/ports/aliases.d/ and hope that jaegar/teK will follow that approach in their reimpl, so it's cleaner and portable between both apps.08:05
Romsterjdk: jre08:05
Romsterinstead of that be /etc/mport/aliases.d/jdk08:05
Romsterand cat /etc/mport/aliases.d/jdk08:05
Romsterthen jdk Pkgfile can include that file in $PKG08:06
Romstermoving that to easiy be done per a port08:06
Romsterthis is more my idea but i think it makes more sense that way.08:06
Romsterany port that has conflicting files with another port and isn't meant to be installed together can list it in that directory.08:08
Romsterdiverse, at this point i'd jsut go with your instinct idea. on my input and stuff can be changed later if need be.08:10
diverseit does sound like there would be fewer operations than having to read line-by-line on just one file for everything08:16
diversewhere as for each port, it's only reading one line, with a list of aliases between spaces.08:18
diversetaking advantage of the fs does sound more optimal08:21
diversesince things would be "prelinked" as files08:22
diverseRomster: although, what would be your opinion on having an `Alias:` tag in the Pkgfiles?08:29
diversearch (/me shivers) already does this as `conflicts=`08:30
diversewhile having separate files in an aliases.d dir would be faster, but at the same time, it sounds redundant to maintain08:32
Romsteraving files in a directory is just a KISS way of doing it.08:33
Romsterwhile keeping Pkgfile creep to a minimum.08:34
Romsterhonestly i been thinking of a group user and alias metatag in Pkgfiles and groups08:34
Romsterbut that's just meh more stuff08:34
diverseso while these ideas are better than the way prt-get does them, I want to also hear how jaeger and teK_ plan to do to solve this in the future (because their reimpls are going to have different names too. I want to be compatible with theirs too)08:37
diversein the meantime I want to implement this later, because I want to dodge prt-get.aliases08:39
diversewell I could implement these now, but I would also feel tied to prt-get.aliases08:41
diversethe locker file sounds pretty straightforward08:43
diverseperhaps I can create a symlink to too08:43
diverseRomster: how about calling the file just 'locker'?08:44
Romsterif you wish but it wont be compatible with prt-get08:44
diverseit's a symlink actually08:44
Romsteroh i guess that'll still work08:45
diverseright, it'll act as a place holder for a less prt-get-ish way of doing things08:45
Romsteri see08:46
diverseand hopefully persuade those 2 to an appname config-neutral way of doing things08:48
diverseRomster: speaking of file I/O, this is going to hurt me pretty badly:
*** heroux has quit IRC09:15
Romsterwhere is the details of the changes?09:16
*** heroux has joined #crux-devel09:17
*** leo-unglaub has joined #crux-devel11:00
*** pitillo has joined #crux-devel11:14
*** leo-unglaub has quit IRC13:33
diverseRomster: would it be simplier just to have a single "lock" command with options "--remove" and "--list" ?14:05
Romsterdiverse, can't see that being any issue14:08
diverseI don't want to create 2 more phantom types with similar functionality, so I think this is the right approach14:09
diverseand I guess I'll have to make the short version of --remove either -e or -m because --root/-r is there14:11
diverseRomster: what looks better to you?14:13
diverse(I'm thinking -m is more mnemonic)14:13
Romsterwell if your going with a meta way each comand can have there own subcommands but i find that anoying.14:14
Romsterjust do as you think is best on that.14:14
diverseI'm not going to implement this right now, I'm going to sleep on it (because I'm tired)14:16
diversejust something to think about14:17
diverseI guess a different terminology is better, perhaps --undo/-u would be better14:21
diverselock --undo <name>14:22
diverseRomster: undo or delete?14:24
diverse(undo is less to type)14:25
diverseGonna to head to bed, night14:28
*** frinnst_ has joined #crux-devel17:08
*** _mavrick61 has joined #crux-devel17:12
*** heroux has quit IRC17:15
*** heroux has joined #crux-devel17:16
*** Feksclaus has quit IRC21:41
teK_jue: maradns needs a bump22:03
*** crash_ has quit IRC23:09
*** crash_ has joined #crux-devel23:10
*** crash_ has quit IRC23:16
*** crash_ has joined #crux-devel23:39

Generated by 2.11.0 by Marius Gedminas - find it at!