Re: Unnecessary change to readline-colored-completion-prefix

[Chet Ramey]

On 10/15/24 2:40 PM, Daniël Gerbrand Haasbroek wrote:

Hi

As stated here 
(https://lists.gnu.org/archive/html/bug-readline/2023-10/msg00025.html),
the custom suffix in `$LS_COLORS` that is used for readline's
"colored-completion-prefix" was "recently" changed to include a
leading dot.  A similar change was not made to the documentation of
readline (and Bash), and this documentation still uses the original

suffix without the leading dot.

This was an oversight, thanks for pointing it out.

 Although the documentation can simply

be changed to use the new suffix, I argue that the correct fix would
be to revert the suffix to its original value (without the leading
dot).

As the original bug report observed, this does not work with at least
some versions of dircolors, so you couldn't specify a value for LS_COLORS
without the dot. That was the entire point of the bug report.

 From here 
(https://lists.gnu.org/archive/html/bug-readline/2023-08/msg00018.html),
the original motivation for adding the leading dot seems to be that

`dircolors` *requires* the file extension to be specified with a leading dot

The existing code, whatever version it was, certainly did. `dircolors'
rejected the suffix without the leading dot.

The original suffix (without the dot) can be easily used with
dircolors by simply using the non-obsolete syntax. Thus, the addition
of the leading dot seems unnecessary, and it breaks all configurations

that use the original suffix.

It appears that there were versions of dircolors out there that didn't
accept the suffix without a dot. If that's the case, working
configurations would have been hit or miss.

 (I could also just be unaware of a

different version of dircolors that does not support this syntax.)

I don't know which version of dircolors the original bug report used.

The code in readline that parses the color specifications and looks
for the custom suffix was contributed; I'm not sure which version of
ls/dircolors/fileutils it came from. It's been there for at least a
decade.

If this is not sufficient to justify reverting the suffix to its
original value, the documentation would need to be changed to use the

new value.

That's a fine suggestion.

--
``The lyf so short, the craft so long to lerne.'' - Chaucer
                 ``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, UTech, CWRU    c...@case.edu    http://tiswww.cwru.edu/~chet/

OpenPGP_signature.asc
Description: OpenPGP digital signature