GNU bug report logs - #61730
30.0.50; Compiler warnings for delq and delete

Previous Next

Package: emacs;

Reported by: Michael Heerdegen <michael_heerdegen <at> web.de>

Date: Thu, 23 Feb 2023 10:30:01 UTC

Severity: wishlist

Found in version 30.0.50

Full log


View this message in rfc822 format

From: Mattias EngdegÄrd <mattias.engdegard <at> gmail.com>
To: Michael Heerdegen <michael_heerdegen <at> web.de>
Cc: 61730 <at> debbugs.gnu.org, Stefan Monnier <monnier <at> iro.umontreal.ca>
Subject: bug#61730: 30.0.50; Compiler warnings for delq and delete
Date: Fri, 24 Feb 2023 14:43:34 +0100
> I think we should add compiler warnings

You rang?

> for `delete' and `delq' function
> calls whose return values are unused.

Let's experiment: warning about for-effect calls to

  mapcar mapcan mapconcat
  delq delete delete-dups delete-consecutive-dups
  cl-delete cl-delete-if cl-delete-if-not cl-delete-duplicates
  sort

results in 34 such calls found on master, most of them about `delq` and `delete`, but `delete-dups`, `cl-delete`, `mapconcat` and `sort` are also represented.

Some of these are no doubt safe, a few of them knowingly so, but it's definitely not obvious from a quick look at the code. It's poor style in any case.

Thus such a warning definitely falls on the beneficial side. Let's do it.

> I think the warnings could be added in a similar way as the "mapcar
> called for effect" warnings work.

That's probably the best place to start (although we prefer warnings to be emitted by the front-end and not in codegen).

We may want to change the warning text from "called for effect" because it's written from the perspective of the compiler; the programmer thinks of it as 'not using the return value'.

For that matter, `with-suppressed-warnings` doesn't work for suppressing this warning very well; we may want to do something about that. This is true for the existing `mapcar` warning as well.

> Adding the same kind of warning for `remq' and `remove' would probably
> also be useful.  This will probably not occur that often but it still
> would be useful I think.

The compiler should already warn about those two since they are declared side-effect-free. Unless somehow `byte-compile-delete-errors` is set during compilation, which can happen if the code messes about with (optimise (safety ...)) -- it's a bit unfortunate. Please tell us if you observe anomalies in this regard.





This bug report was last modified 1 year and 286 days ago.

Previous Next


GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.