Date: Wed, 10 Dec 1997 11:12:30 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Bill Currie <bcurrie AT tssc DOT co DOT nz>
cc: djgpp-workers AT delorie DOT com
Subject: Re: Possible enhancements on v2.02
In-Reply-To: <348D942B.2422@tssc.co.nz>
Message-ID: <Pine.SUN.3.91.971210111143.10111B-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


On Wed, 10 Dec 1997, Bill Currie wrote:

> Isn't that a bug? If LFN=n then the program should never see long file
> names.  It seems you're saying that no matter what LFN is set to, you
> will always get the long fine names on the command line.

If LFN is only set on DJGPP.ENV, then you always get the long names,
that's true.  And yes, I think it's a bug.  It will rear its ugly head
when you have NameNumericTail set to a non-zero value (if it's zero,
the probability of trouble is much lower, but it still exists).

The problem is that fixing this is not easy.  DJGPP.ENV cannot be read
before processing at least part of the command line (you need to know
what argv[0] is, to get to the appropriate section of DJGPP.ENV), so
this means either some serious changes in the way command line
globbing works (divide it into two parts) or we need to do it again if
DJGPP.ENV resets LFN.

I've asked for opinions here about a week ago, but didn't get any.  I
guess people are unwilling to introduce non-trivial changes into the
startup code for a minor bug-fix release such as v2.02, since the
current code took a lot of iterations to get right, and since nobody as 
yet has complained about this.

If there's anybody else (besides Bill, that is) who thinks this should
be corrected, or has ideas on how to do that, please speak up.