BluesKaj | Hi folks | 12:33 |
---|---|---|
santa_ | hi everyone | 12:56 |
santa_ | RikMills: if we have fw 5.82 pre-release this weekend I would like to do it | 12:57 |
santa_ | I have the initial version of the watch file parsing code to test | 12:58 |
RikMills | santa_: as long as it worked on straight away, that is fine | 12:58 |
RikMills | I want to be able to test build new plasma with it ASAP | 12:59 |
RikMills | e.g. make a dummy release set of plasma tars | 12:59 |
RikMills | that way will not have to wait for the beta to get much of the work prepared | 13:00 |
santa_ | sure, I have just one commit more to push to master and then would be actually ready to test | 13:01 |
RikMills | cool | 13:01 |
santa_ | RikMills: ok, so last commit is already there, pushed to master, if may steal you a few mins, let me explain what we have now: | 14:22 |
santa_ | - we have a new KAWatchFile class | 14:22 |
santa_ | - we have a new debug program named parse-watch-file | 14:22 |
santa_ | - and we have something different in the package-info debug program output | 14:24 |
santa_ | so ... | 14:25 |
santa_ | if you execute parse-watch-file against any pkg clone you will see at the end "Tarball name patterns: [...]" | 14:25 |
RikMills | right | 14:26 |
santa_ | these are all the possible tarball names for the package which may be under /home/user/kde-ftp/... | 14:26 |
santa_ | if you execute package info, you will see in the middle of the stdout something like "Upstream tarball paths: [...]" | 14:27 |
santa_ | these are all the possible upstream tarball full paths | 14:28 |
santa_ | so the code we have now, when doing a symlink in build-area check sequentially all this paths and creates the symlink targeting the first existing one | 14:29 |
santa_ | this way, what we have now it's better than what we use to have: previously the tarball name was just built like <kde-ftp_path>/<tarball_name>-<upstream_tar_version>.tar.xz | 14:30 |
santa_ | and of course only .tar.xz was supported | 14:31 |
RikMills | sounds good/better :) | 14:31 |
santa_ | now we have a better set of guesses | 14:31 |
santa_ | if you see the code of the KAWatchFile, now you will understand why I said some time ago I wanted to move the code responsible to find out the tarball name to a separate class :) | 14:32 |
santa_ | (that would be difficult to implement in a single function inside KASrcPkg) | 14:33 |
santa_ | more stuff & limitations: | 14:33 |
santa_ | about the kdsoap an sddm watch files, they have some "challeging" things that the fw/plasma/apps don't have: | 14:34 |
santa_ | - non .tar.xz extension | 14:34 |
santa_ | - lines split into multiple lines with final '\' | 14:35 |
santa_ | - sddm also has this filenamemangle thing, which makes things more complicated because it forces you to parse the "opts=" things and try to replicate the "s/..." | 14:36 |
santa_ | that being said, while I expect this new code to be better than the previous one, I don't expect it to be infallible: | 14:37 |
santa_ | - if you read the uscan manpage and the KAWatchFile code and understand it, you will find out what I already know: some watch files might not work | 14:38 |
santa_ | - also you have to take into account that what KAWatchFile does it's just a *guess* of the tarball name that uscan is going to download, but you never know with 100% accuracy | 14:39 |
RikMills | fair enough. thanks! | 14:39 |
santa_ | so, last, but not least: for now I plan to keep the code like, but if you hit any watch file that doesn't work let me know, so we could either: | 14:40 |
santa_ | a) improve the guess so it works better | 14:40 |
santa_ | or | 14:40 |
santa_ | b) provide an overrides system indexed by source package name in ka-metadata | 14:41 |
RikMills | will do | 14:41 |
santa_ | for now I would like to keep the code mostly as it is, and explore the a) option | 14:41 |
santa_ | and that's it, thanks for your time | 14:42 |
RikMills | great, and thank you | 14:44 |
santa_ | yw | 14:44 |
mparillo | ii ISOs have landed: http://cdimage.ubuntu.com/kubuntu/daily-live/current/ Installed just fine to a VM. | 15:06 |
RikMills | \o/ | 15:09 |
RikMills | santa_: https://mail.kde.org/pipermail/release-team/2021-May/012288.html | 15:20 |
santa_ | RikMills: nice! I'm going to start then if you don't mind | 15:21 |
RikMills | please do :) | 15:21 |
BluesKaj | so far so good on 21.10 Impish :-) | 15:26 |
RikMills | plasma-wayland-protocols 1.3.0 uploaded | 16:32 |
RikMills | santa_: ^ which has a normal tar name this time ;) | 16:33 |
krytarik | RikMills: https://code.launchpad.net/~krytarik/ubuntu-cdimage/flavor-assets/+merge/401751 - now that I believe I've met all of vorlon's requirements, would you mind fulfilling the team repo one? | 16:41 |
RikMills | krytarik: appears I have no permissions on the ~kubuntu-website team | 16:54 |
RikMills | looks like mparillo and @Sick_Rimmit do | 16:55 |
RikMills | https://launchpad.net/~kubuntu-website/+members#active | 16:55 |
krytarik | Oh indeed, I was mistaking Rick Timmis for you! >_< | 17:02 |
santa_ | typical mistake | 17:03 |
santa_ | so many Ri[c]ks in the room | 17:03 |
mparillo | The owner of the team is https://launchpad.net/~ryanakca | 17:58 |
mparillo | And Rick (https://launchpad.net/~rick-timmis) is one of the Admins. | 18:03 |
krytarik | I doesn't matter much really though who is owner or admin on the team - anybody on it should be able to create a new repo under it. | 18:05 |
krytarik | Though it's mildly hilarious how much cruft there is on some teams.. >_> | 18:09 |
RikMills | konsole with the horrible new toolbar now landing in impish :S | 21:40 |
IrcsomeBot | <X> (Photo, 610x357) https://irc-attachments.kde.org/OmyMrJt1/file_43621.jpg No toolbar Konsole. 10/10. Would recommend | 21:44 |
mamarley | LucKily, as with other things KDE, it is Konfigurable enough to disable the silly toolbar! | 22:29 |
valorie | heh | 22:33 |
IrcsomeBot | * DarinMiller prefers konsoie with no menu or title bar | 22:57 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!