[tex-live] Not: Kpathsea in TL2007

Philip Taylor (Webmaster) P.Taylor at Rhul.Ac.Uk
Mon Feb 19 13:37:16 CET 2007

Reinhard Kotucha wrote:

 > If I say "Readme" I mean "Readme" and if I say "Readme." I mean
 > "Readme.".  It's simply wrong if I get "Readme" when I say "Readme.".

If you believe that the period is a part of the filename, true;
if you believe that it is a delimiter, false.

 >   > Yes, all of these apart from "Makefile" would benefit enormously
 >   > from having a ".txt" or ".log" extension.
 > No, nothing/nobody will benefit.

The benefit is that it is immediately apparent what sort of file
it is; as it stands, one has to guess from the filename, which
is a very poor choice of heuristic, IMHO.

  Extensions are not unique and I
 > guess that Perl is unable to process a file like "uplr8t.pl".

That would depend on its contents.  If it contained a TeX font
property list, no. If it contained a Prologue source file, no.
But if it contained a PERL program, then yes.  Of course, it is
a sad fact of life that whilst we use very short extensions (two
or three characters), the probability of semantic overload is
very high : what would be nice is if file systems evolved such
that the /type/ of a file was stored in the file descriptor, so
that when one performed a directory listing one was told not only
the name and extension but also its type (Prologue source file, TeX
Property list, PERL source file and so on).
 >   > And if only "Make" would expect files of type ".mak", then we
 >   > could do away with extensionless files for ever :-)
 > Sigh...  Why do you suggest ".mak" and not ".make"?

Just to stick with the current received wisdom that, in general,
three characters are necessary and sufficient.  In fact, all of
my PERL programs have extension ".perl" on my own machine; the
example I referred to previously was on someone else's machine.

 > It's good that you like Windows so much but you can't expect that
 > UNIX users like all these restrictions.

But I'm not proposing /restrictions/ : I'm proposing /improvements/ ...

 > You cannot force UNIX users to live in the stone-age.

That's the very age I'm trying to move them out of :-)

Neanderthal man was moderately successful, but was eclipsed
by Cro-Magnon man : Unix was moderately successful (for its
time), but its time is long since past and it's time for
Unix users to move on !

** Phil.

More information about the tex-live mailing list