pacaur - an AUR helper that minimizes user interaction

willemw · 2013-04-14 16:42:27

What do you think of the following enhancements for pacaur:
- display aur package comments
- configurable "continue building" prompt after viewing the final PKGBUILD file

Spyhawk · 2013-04-14 16:57:31

willemw wrote:

- display aur package comments

Bloated. I'd prefer to see an external tool doing that. Adding a RSS feed to each package page might help, as currently the whole HTML code must be parsed.

willemw wrote:

- configurable "continue building" prompt after viewing the final PKGBUILD file

This could be a very neat addition.

Last edited by Spyhawk (2013-04-14 16:58:23)

willemw · 2013-04-14 18:12:46

Displaying aur package comments would add a lot of value though to the already efficient workflow (i.e. inspecting all packages before building and installing them).

Spyhawk · 2013-04-14 18:43:27

willemw wrote:

Displaying aur package comments would add a lot of value though to the already efficient workflow (i.e. inspecting all packages before building and installing them).

You don't need to view comments for each and every packages you install, but only when a problem arises (ie, package not compiling). However, I do agree that there is currently a lack of tool to do that. Could be a very useful addition to aurtomatic for example (but there is no neat way to parse and retrieve the comments).

willemw · 2013-04-15 08:44:58

There are so many ways and tools/libraries to extract aur comments. For example, have lynx as an optdeps and conditionally do something like this (can be configurable and/or prompt the user):

lynx -dump https://aur.archlinux.org/packages/pacaur/?comments=all | grep --max-count=5 --after-context=99 --color=never "Comment by"

I don't find this example at all bloated.

Spyhawk · 2013-04-15 10:17:20

Thanks for your input. Retrieving comments seems much more simple that I initially thought (yaourt does it in a much complicated way). I'm still not convinced that should be part of pacaur though, but I'll have a look at it as a possible new (optional) feature.

Last edited by Spyhawk (2013-04-15 10:17:30)

willemw · 2013-04-15 11:14:30

And yaourt sorts the comments the "wrong" way. Irritating.

To stop extracting text below the initial aur comment, do for example something as simple as:

lynx -dump https://aur.archlinux.org/packages/pacaur/?comments=all | grep "AUR v" --before-context=99 | grep --max-count=5 --after-context=99 --color=never "Comment by"

Or use another utility (sed, awk, ...).

Showing additional information (coming from "cower -i"), possibly only in combination with displaying aur comments, could be: "Submitted" or "Last Modified", "Out of Date", "Orphaned".

narco · 2013-04-16 10:10:46

Hi, when I launch "pacaur -Syu" I have these warnings messages:

:: initscripts is not present in AUR -- skipping
:: leed-svn is not present in AUR -- skipping
:: libdrm-nouveau1 is not present in AUR -- skipping

why?
thanks

Last edited by narco (2013-04-16 10:11:59)

Spyhawk · 2013-04-16 10:19:56

narco wrote:

Hi, when I launch "pacaur -Syu" I have these warnings messages:
why?
thanks

pacaur warns about "truely foreign" packages: theses are not in the repo, nor in the AUR. Most of the time, these are repository packages that have been superseded by other packages and are now obsolete (ie, initscripts has been superseded by systemd).

narco · 2013-04-16 14:32:21

thanks, so is there a way to hide these messages?

bananagranola · 2013-04-16 15:14:26

You might be better served looking up and replacing those packages. For example, initscripts -> systemd, libdrm-nouveau -> libdrm. I don't know what leeds is.

Last edited by bananagranola (2013-04-16 15:17:20)

Spyhawk · 2013-04-16 15:28:19

leed-svn seems to have been replaced in the AUR by leed-git.

narco · 2013-04-16 15:28:55

 core/systemd 201-1 [installed]
    system and service manager
extra/libdrm 2.4.43-1 [installed]
    Userspace interface to kernel DRM services

and Leed (Light Feed) is a free RSS aggregator that I have tried a long time ago (in the aur now there's leed-git)

Spyhawk · 2013-05-09 09:33:15

DaveCode> Your email doesn't seem to be valid (my answer mail get rejected by the server), so here is the answer to your PM:

Thanks for pacaur - it's the greatest thing.
We could not install tor-browser-en without its Java stuff. All that Java should be optional; Maxr hardwired it needlessly.
I'd like pacaur to recheck all deps AFTER the PKGBUILD file is edited. But still, it could be a "bug" that pacaur ignores pacman's --nodeps option.
So things to think about....no reply please, just a note here, and thanks for everything you do.

Hi DaveCode,

By design, pacaur doesn't recompute the dependency tree after a PKGBUILD has been edited. The reasons for this are that the PKGBUILD is assumed to be correct, and if there is an error the PKGBUILD should be corrected by the maintainer instead of adding lot of unnecessary complexity in the code.

In this case, the java-runtime deps should be made optional by the maintainer, as noted in the comments. As for the "--nodeps" option, a PKGBUILD (assuming it is correctly written) couldn't be compiled or couldn't be launched without its makedepends/dependencies, so it simply doesn't make sense.

And if you need to tinker the PKGBUILD for any reason, simply use cower to download the PKGBUILD (cower -d pkg), edit it and run makepkg to compile the package. Correcting the PKGBUILD upstream should however be the top priority.

vri · 2013-05-18 10:42:22

I use pacaur a lot without problems generally, so thanks for this great helper.

Now when I try to install the AUR package python-pyramid, it does not seem to get past ':: resolving dependencies...'; I waited for quite some time, but it still was busy.

All dependencies are installed on my system.

Any idea what may be the problem?

Cheers, Theo

Spyhawk · 2013-05-18 11:27:31

vri> Thanks for the report, I'll have a look asap.

Edit: The sanity function done before the sudo security check was going into an infinite loop, but this bug was triggered only in some rare situation. Fixed in 4.1.8 and I also enhanced robustness of that part of the code.

Last edited by Spyhawk (2013-05-18 14:40:26)

vri · 2013-05-19 12:24:57

Thanks for the speedy follow up! Works fine now.

darwell.j · 2013-05-23 14:15:52

Hey guys,

pacaur seems to stuck when using AUR operations. gpg2keys_hkp uses 100% CPU while it's stuck. If I leave it for a few hours, the operation continues.
Here's the output of bash -x pacaur -Sua : http://pastebin.com/raw.php?i=Hw0TSvQJ

Spyhawk · 2013-05-23 14:55:18

darwell.j > Can't reproduce here. Does running "expac -S -1 -v %n android-sdk" generate any output in the shell?

Last edited by Spyhawk (2013-05-23 14:55:40)

darwell.j · 2013-05-23 16:13:35

No, just hangs there...
[EDIT]. Running "expac" does the same - it just hangs there while gpg2keys_hkp uses 100% CPU. I guess I have to look for the problem somewhere else...
[EDIT2]. Disabling xorg113 and catalyst repos seems to solve the problem. Something must be wrong with these repos...

Last edited by darwell.j (2013-05-23 16:20:02)

Spyhawk · 2013-05-23 16:35:24

darwell.j wrote:

[EDIT2]. Disabling xorg113 and catalyst repos seems to solve the problem. Something must be wrong with these repos...

Third party binary repo usually bring more problems than they solves issues. Better avoid them and use the AUR instead imho.
Edit: Seems there is an issue with the database signature of that repo. See this expac issue.

Last edited by Spyhawk (2013-05-23 16:54:29)

DaveCode · 2013-05-25 06:37:25

Spyhawk wrote:

pacaur doesn't recompute...after...edited...
deps should be made optional by the maintainer...
"--nodeps"...a PKGBUILD...couldn't be compiled...
if you need to tinker...use cower

If not for tinkering, why does pacaur prompt to edit the PKGBUILD? Just provide a viewer (less).

Re those correctness assumptions - AUR is more brittle/broken than official repos - so they don't always hold.

Going from memory, we felt pacaur was wrong to omit --nodeps. We love pacaur for only lightly wrapping pacman. a superset. Here it's a subset. Why might pacman have a --nodeps flag?

Any admin editing a PKGBUILD and calling --nodeps should be allowed to crash. Many AUR packages fail to build as things stand, even with good PKGBUILDs. So crashing AUR is nothing new.

Staff wanted to remove Java deps in pacaur's edit screen. Ultimately yes, an AUR maintainer should (and did) do it. We KNEW what to do, but pacaur stopped us.

Hey, another topic while I'm here....What's the official way to run pacaur -Syua --noconfirm --noedit from a nightly cron? Yes, we know it's not kosher, but nights are good times to build stuff.

Thanks again! We do love pacaur.

Spyhawk · 2013-05-25 11:18:09

DaveCode wrote:

If not for tinkering, why does pacaur prompt to edit the PKGBUILD? Just provide a viewer (less).

It does prompt to "view" the PKGBUILD. Just use 'editor="less"' in your config file.

If needed, everything else can be edited, but this is just a consequence of using makepkg (only AUR dependencies changes aren't taken into account). I never expected the user to change the dependencies, and pacaur assumes they are correct.

DaveCode wrote:

Going from memory, we felt pacaur was wrong to omit --nodeps. We love pacaur for only lightly wrapping pacman. a superset. Here it's a subset. Why might pacman have a --nodeps flag?
Any admin editing a PKGBUILD and calling --nodeps should be allowed to crash. Many AUR packages fail to build as things stand, even with good PKGBUILDs. So crashing AUR is nothing new.

In practice, "--nodeps" would simply break the build. The only reason you wanted --nodeps was because the PKGBUILD was incorrectly written, but I see your argument for consistency here (which makes sense). I wanted to implement a more general option to pass any option to makepkg, but I'll add some code to pass this option directly to makepkg.

DaveCode wrote:

Staff wanted to remove Java deps in pacaur's edit screen. Ultimately yes, an AUR maintainer should (and did) do it. We KNEW what to do, but pacaur stopped us.

Just to be clear: I don't believe it is pacaur's job to fix what should be fixed upstream. If there is an issue with a PKGBUILD, the user should ask the maintainer to fix it. Yes, I know, it's slower, it takes time, but it is also a way to improve the AUR for everybody.

DaveCode wrote:

Hey, another topic while I'm here....What's the official way to run pacaur -Syua --noconfirm --noedit from a nightly cron? Yes, we know it's not kosher, but nights are good times to build stuff.

There is no official way, since pacaur has never been designed to be used in a cron job... But well... have a look at the wiki.

Last edited by Spyhawk (2013-05-25 11:38:32)

donbex · 2013-05-30 07:47:11

First of all, thanks for coding pacaur, which has been my favourite AUR helper since the late bauerbill.
I'm trying to install the xmp package from the AUR, but somehow pacaur looks for the wrong package:

$ pacaur -v
pacaur 4.1.8

$ pacaur -S xmp 
:: Package(s) xmp not found in repositories, trying AUR...
:: resolving dependencies...
:: no results found for bmp

$ pacaur -Ss xmp
...
aur/xmp 4.0.6-1 (77)
    The Extended Module Player is a portable command-line module player that supports over 90 module formats (Amiga, Atari, ..)
...

Spyhawk · 2013-05-30 08:00:33

donbex> bmp is a makedependency of libxmp, which is in turn a dependency of xmp. The package 'bmp' doesn't exist, so pacaur can't find it. The PKGBUILD of libxmp must be corrected here. The package 'bmp' might have been deleted/replaced over time.

But yes, I agree that the error output isn't very explicit... more like a "WTF!?" situation

Last edited by Spyhawk (2013-05-30 08:04:02)

Arch Linux

#351 2013-04-14 16:42:27

Re: pacaur - an AUR helper that minimizes user interaction

#352 2013-04-14 16:57:31

Re: pacaur - an AUR helper that minimizes user interaction

#353 2013-04-14 18:12:46

Re: pacaur - an AUR helper that minimizes user interaction

#354 2013-04-14 18:43:27

Re: pacaur - an AUR helper that minimizes user interaction

#355 2013-04-15 08:44:58

Re: pacaur - an AUR helper that minimizes user interaction

#356 2013-04-15 10:17:20

Re: pacaur - an AUR helper that minimizes user interaction

#357 2013-04-15 11:14:30

Re: pacaur - an AUR helper that minimizes user interaction

#358 2013-04-16 10:10:46

Re: pacaur - an AUR helper that minimizes user interaction

#359 2013-04-16 10:19:56

Re: pacaur - an AUR helper that minimizes user interaction

#360 2013-04-16 14:32:21

Re: pacaur - an AUR helper that minimizes user interaction

#361 2013-04-16 15:14:26

Re: pacaur - an AUR helper that minimizes user interaction

#362 2013-04-16 15:28:19

Re: pacaur - an AUR helper that minimizes user interaction

#363 2013-04-16 15:28:55

Re: pacaur - an AUR helper that minimizes user interaction

#364 2013-05-09 09:33:15

Re: pacaur - an AUR helper that minimizes user interaction

#365 2013-05-18 10:42:22

Re: pacaur - an AUR helper that minimizes user interaction

#366 2013-05-18 11:27:31

Re: pacaur - an AUR helper that minimizes user interaction

#367 2013-05-19 12:24:57

Re: pacaur - an AUR helper that minimizes user interaction

#368 2013-05-23 14:15:52

Re: pacaur - an AUR helper that minimizes user interaction

#369 2013-05-23 14:55:18

Re: pacaur - an AUR helper that minimizes user interaction

#370 2013-05-23 16:13:35

Re: pacaur - an AUR helper that minimizes user interaction

#371 2013-05-23 16:35:24

Re: pacaur - an AUR helper that minimizes user interaction

#372 2013-05-25 06:37:25

Re: pacaur - an AUR helper that minimizes user interaction

#373 2013-05-25 11:18:09

Re: pacaur - an AUR helper that minimizes user interaction

#374 2013-05-30 07:47:11

Re: pacaur - an AUR helper that minimizes user interaction

#375 2013-05-30 08:00:33

Re: pacaur - an AUR helper that minimizes user interaction

Board footer