guix-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[bug#45984] [PATCH 0/5] Fix recursive importers


From: Ludovic Courtès
Subject: [bug#45984] [PATCH 0/5] Fix recursive importers
Date: Tue, 26 Jan 2021 23:24:56 +0100
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/27.1 (gnu/linux)

Hi!

zimoun <zimon.toutoune@gmail.com> skribis:

> This corner case #1 happens when the package does not exist; then the function
> 'lookup-node' is not able to "unpack" the 'values' and throw and ugly
> backtrace, as exposed in bug#44114 <http://issues.guix.gnu.org/44115#3>.
>
> With these trivial patches, it is fixed for all the importers except 'opam'
> (because of 'and-let' which needs some care).

Neat!

> Now, instead of throwing an ugly backtrace, it simply say almost nothing:
>
> $ ./pre-inst-env guix import cran do-not-exist -r
> error: failed to retrieve package information from 
> "https://cran.r-project.org/web/packages/do-not-exist/DESCRIPTION": 404 ("Not 
> Found")
>
> $ ./pre-inst-env guix import pypi do-not-exist -r
> following redirection to `https://pypi.org/pypi/do-not-exist/json/'...
> #f
>
>
> This non-existent message is because the error is poorly handled.  With the 4
> patches, the situation is the same as "guix import gem" for all the importers
> with the recursive option.  One way for a better error handling is done in the
> last commit for 'guix import gem' only; the same trick can be done for all.
>
> $ guix import gem do-not-exist -r
> #f
>
> $ ./pre-inst-env guix import gem do-not-exist -r
> guix import: error: failed to download meta-data for package 'do-not-exist'

I think we do want this error message.  Why should we ignore
non-existent packages when doing ‘-r’?  It would think they’re still a
problem, no?

> In my opinions, UI messages should not appear in guix/import/*.scm but only in
> guix/scripts/*.scm.

I agree with the general idea, though sometimes taking this shortcut is
beneficial (maybe not in this case?).

> If I understand correctly, then the way the errors are reported could be
> uniformized between all the importers, and maybe the snippet in the 
> subcommands
> "guix import <from>" could be refactorized a bit.

Probably.  ‘-r’ started as an option specific to one importer, but now
that most of them (?) support it, it’d make sense to rethink the
interfaces.

Thanks for looking into it!

Ludo’.





reply via email to

[Prev in Thread] Current Thread [Next in Thread]