Mail Archives: djgpp/1998/08/30/14:15:44
Eli Zaretskii <eliz AT is DOT elta DOT co DOT il> wrote:
>On Sun, 30 Aug 1998, Peter J. Farley III wrote:
>
>> because a relative path (../intl/lib*) works, but the
>> absolute path version "//h/mc-4.1.35/intl/lib*" causes the "ls"
>> command to produce ENOENT errors.
>
>You should *never* let DJGPP programs except Bash itself see pathnames
>such as //h/mc-4.1.35/. The only program that understands them is
>Bash itself; other DJGPP programs will choke on them.
Understood. I'll have to adjust the makefile, then.
>I don't have enough information to tell you how to correct that in
>your case. If you run GCC from the Bash prompt, just use the usual
>DOS h:/mc-1.4.35 form. If the //h/foo pathnames come from the
>Makefile, you will need to figure out how did they get there and
>change the configure script accordingly.
>
>One thing I always do when running configure scripts is to set
>PATH_EXPAND=y in the environment; you should also do that when running
>Bash interactively. Usually, this does the trick for me.
Yes, I have PATH_EXPAND=y in the environment; I set it in the BAT file
I use to invoke bash, but thanks for reminding me to check.
>I tried to ask the other DJGPP developers whether it was a good idea
>to add in v2.02 support for the //h/foo pathnames in the core DJGPP
>library (so that all DJGPP programs will know about that), but there
>were too many negative responces with concerns about things that might
>break, so I guess it won't happen.
That's too bad. Is your question and their responses available in
DJ's archive of the mailing list? I'd like to go read about the
things that might break.
Thanks as usual for the quick, cogent reply.
----------------------------------------------------
Peter J. Farley III (pjfarley AT nospam DOT dorsai DOT org OR
pjfarley AT nospam DOT banet DOT net)
- Raw text -