Mail Archives: djgpp/2009/06/30/18:30:48
Hi,
On Jun 30, 1:05=A0pm, Eli Zaretskii <e DOT DOT DOT AT gnu DOT org> wrote:
>
> I don't see how could letter-case matter here. =A0If you can show me
> otherwise, please do. =A0The exact error message would be nice.
Well, it seemed to work with 4DOS without any modifications, so I'm
blindly guessing that case matters. Below is the exact error message:
-------------------------------
[ FreeDOS ] Mon 06-29-2009>make
cd lib-src
d:/gcc/bin/make.exe top_srcdir=3De:/emacs version=3D"23.0.95"
make.exe[1]: Entering directory `e:/emacs/lib-src'
make.exe[1]: *** No rule to make target `/test-distrib.c', needed by
`test-distrib'. Stop.
make.exe[1]: Leaving directory `e:/emacs/lib-src'
make.exe: *** [lib-src] Error 2
-------------------------------
> I suspect that the problem could be not with top_srcdir in the
> top-level Makefile, but rather with this line from lib-src/Makefile:
>
> =A0 =A0 srcdir :=3D $(subst \,/,$(shell command.com /c cd))
I changed both Makefile and lib-src/Makefile, both top_srcdir and
srcdir, respectively.
> Do you have command.com on FreeDOS, and if you do, what does it do
> with this line? =A0What does $(srcdir) get set to?
FreeCOM is named COMMAND.COM for compatibility. I guess it gets set to
"E:\EMACS\lib-src". (I should really test again and use djecho to
verify it.) All I'm saying is that it won't work with FreeCOM unless I
manually change it. I can't say why.
> > Wow, the smallest upgrade in the history of the universe. Taking a big
> > risk, eh? Living on the bleeding edge? =A0 ;-) =A0 =A0j/k
>
> It's very simple: the only other port is of GCC 4.x, and I don't want
> yet to go there.
Hey, it's fine, I use all kinds of GCCs, so it's no biggie. In fact, I
think the AVR targeting ones work better in 3.x than 4.x. Latest
OpenBSD 4.5 still uses 3.3.5 (I think some non-x86 targets (Sparc??
VAX??) have bugs with newer versions). However, most other people use
4.x due to its improvements (although *BSD avoid anything newer than
4.2.1, if even, due to GPLv3). Actually, I think 4.0.x was vaguely
buggy, 4.2.x had a big unknown slowdown, and 4.3.2 brought things back
up to par. That's not to say they don't whine a lot (-Wno-parentheses
is pretty much required, IMHO). But hey, 3.4.4 is from 2005, so it's
not old, and I still use it (and older!) on various computers, esp. my
P166 (although in all honesty, despite being twice as fast as 4.4.0,
it's twice as slow as 2.95.3, go figure). For laughs you may want to
try the DJGPP/ELF hack (COFF or ELF supported), which uses 4.0.1. GCC
4.x really does optimize better (esp. C++) although not a billion
times better or anything.
- Raw text -