[biblio] Relative paths and BIBINPUTS

[Tohiko Looka]

Hello Karl, Thanks for your answer.

Yes, that does explain the behaviour. I didn't know that shell treated
"../" and "./" paths this way.
I just checked and \input with TEXINPUTS have a consistent behaviour.

Cheers, Tohiko.

On Mon, Jun 15, 2020 at 9:40 PM Karl Berry <karl at freefriends.org> wrote:

> Hi Tohiko - I admit I did not try to reproduce your problem (thanks for
> the careful report and scripts; much appreciated), because I saw this in
> your mail:
>
>     only when the path starts with `..` (or `.`).
>
> I made the decision when I wrote kpathsea that if a filename started
> with an explicit "." or "..", it would be treated as an absolute path
> (same as if it started with "/" or "C:" or whatever), and only looked up
> as-is, not along the search paths.  Normal shells do it that way, so it
> seemed like this was the most intuitive thing to do.
>
> I wouldn't claim that the entire TeX suite is 100% consistent in this
> regard, but I think that might explain the behavior you're seeing?
> --thanks, karl.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/biblio/attachments/20200615/c1683cad/attachment.html>