bug#79445: `ln -s` should raise a warning about pwd-relative vs symlink-location-relative targets

PS. This is about more than just learnability for newbies -- this issue
tripped me up after 18 years of using Linux as a desktop and/or dev
environment, because I ran into the rare case where a program (in this case
https://github.com/bbuhrow/yafu/) needed a separate working directory for
each instance and was useful to run multiple instances of in parallel (in
this case, factoring multiple different composites for https://factordb.com/
).

Sincerely,
Chris Hennick


On Sun, Sep 14, 2025 at 2:15 AM Chris <[email protected]> wrote:

> Isn't it better to surprise users who know what they're doing with a
> warning, than to surprise users who *don't* know what they're doing with
> the lack of one?
>
> On Sat, Sep 13, 2025 at 11:54 PM Collin Funk <[email protected]>
> wrote:
>
>> Paul Eggert <[email protected]> writes:
>>
>> > On 2025-09-13 03:16, Chris wrote:
>> >> It seems to me it should be easy enough to alert users to this gotcha
>> by
>> >> printing a warning to stderr when creating a symlink
>> >
>> > I dunno, that gotcha has been present in Unix and Linux for nearly 50
>> > years now, and lots of people are used to the gotcha would plausibly
>> > object to a warning.
>>
>> FWIW, relative symbolic links and dangling symlinks are covered in the
>> manual. You can read it online [1], or using the following command in
>> your terminal:
>>
>>     $ info '(coreutils) ln invocation'
>>
>> I agree that it probably is a point of confusion for someones first
>> encountering symbolic links, but it is a perfectly valid use of them. So
>> I think emitting a warning there would cause some complaints.
>>
>> New warnings tend to surprise people. I'm sure Paul remembers the many
>> long threads about the warning that 'egrep' and 'fgrep' are obsolete. :)
>>
>> Collin
>>
>> [1]
>> https://www.gnu.org/software/coreutils/manual/html_node/ln-invocation.html#ln-invocation
>>
>