Make problem in gcc-2.95.2-msvcrt?

[Joseph M. Hinkle]

Recent installation of gcc-2.95.2 went smoothly and a number of test
programs compiled correctly.

THEN I discovered that 'make' was a soft link left over from the Cygwin
installation /bin/make -> cygwin-b20/H-i586-cygwin32/bin/make

That's the older gcc-2.95 installation.

When the link was removed so each installation used its own 'make', the one
in gcc-2.95.2 failed miserably:  A simple target yielded this:

$ make clean

rm -f test.o test.exe
process_begin: CreateProcess(E:\CygnusRoot\bin\rm, rm -f test.o test.exe,
..) failed
make (e=193): %1 is not a valid Win32 application

It would be easy to think this is some installation fumble, but 'rm -f
test.o' works just fine, even specificially E:/CygnusRoot/bin/rm -f test.o
works fine.  (All the utilities in /bin are soft links to the binaries in
the Cygnus installation that aren't present in gcc-2.95.2, there are no
soft links to the compiler routines)

Now c++ works well, and that means gcc, and ld, and a compilation proceeds
properly, but there's these little shell snags, I noticed a line in a
makefile TOP = $(shell pwd) also choked the new 'make'

Has anyone else noticed this behaviour? Could someone post a clue here on
troubleshooting this?
And while I'm at it, bash has always been a little funny, when editing a
long line called from the history it responds very slowly to backspace
keypresses, such that a repeating key will cause the insertion of [D's and
eventually lock it up.  I understand there's some problem in bash that way,
is anyone working on that? (I think of looking into fixing things in my
spare time) ....

Sorry about the phony email address, the junk mail is getting horrendous.
I'll check in here for any possible responses.

jmh



--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com