Subject: Porting g++ 1.40.3
Date: Thu, 26 Dec 91 04:02:45 CET
From: Wolfgang Thiel < UPSYF108@comparex.hrz.uni-bielefeld.de>
To: linux-activists@joker.cs.hut.fi
Hi,
Did somebody else try to port g++-1.40.3 to linux? I am having problems
with the estdio FILE structure. I tried the MASSCOMP define, setting __rend
and __rptr to something......., but - when trying to build libg++-1.39.0 -
I always get errors in PlotFile.h; this has to do something with those
inline functions, I'm shure. (And the problem with estdio only starts
AFTER that!!). So, questions:
1. Who has come a step further ?
2. Who has - PD - sources for a USG compatible stdio library ?
(I remember some PD source for some UNIX shell which included a stdio
library; but I cannot remmeber the name....).
Wolfgang
Please reply to upsyf108@comparex.hrz.uni-bielefeld.de, I cannot receive
the list at the moment: this wonderful vm/cms kermit is trying to send
from my usual account upsyf173, and I cannot login again to read the mail.
Subject: Re: Porting g++ 1.40.3
To: Wolfgang Thiel <UPSYF108@comparex.hrz.uni-bielefeld.de>,
In-Reply-To: Your message of Thu, 26 Dec 91 04:02:45 +0700.
Date: Wed, 01 Jan 92 13:14:17 MST
From: drew@kinglear.cs.Colorado.EDU
--------
Hi,
Did somebody else try to port g++-1.40.3 to linux? I am having problems
with the estdio FILE structure. I tried the MASSCOMP define, setting __rend
--------
Gcc 2.0 is imminient. Some one in our CS department is playing with a beta
version, and although I'm unsure of the release date, it may be installed
before next semester. Gcc and g++ also become one product in 2.0.
Save the trouble of porting another gcc in the future, and do it when 2.0
is available. Then send in diffs to the config files to the gnu people,
any conditionaly compiled code in any source files (should be
unecessary) and have it part of the standard gcc distribution.
Subject: gcc 2.0
Date: Thu, 2 Jan 1992 01:47:37 +0200
From: Linus Benedict Torvalds <torvalds@cc.helsinki.fi>
To: Linux-activists@joker.cs.hut.fi
drew@kinglear.cs.Colorado.EDU: "Re: Porting g++ 1.40.3" (Jan 1, 13:14):
>
> Gcc 2.0 is imminient. Some one in our CS department is playing with a beta
> version, and although I'm unsure of the release date, it may be installed
> before next semester. Gcc and g++ also become one product in 2.0.
>
> Save the trouble of porting another gcc in the future, and do it when 2.0
> is available. Then send in diffs to the config files to the gnu people,
> any conditionaly compiled code in any source files (should be
> unecessary) and have it part of the standard gcc distribution.
I've also been fooling around a bit with a beta version of gcc-2.0 -
that's the reason I wanted to have some kind of 387-emulation in 0.12.
I can happily report that cc1 compiles without changes, but the version
I compiled sometimes aborts without any error message (not due to linux:
it does the same thing on a Sun). In any case it seems that with the
387 emulation in place, gcc-2.0 shouldn't need any difficult porting: it
compiled at once after "./configure --gas i386-gas-sysv", and after I
removed a "-lPW" and a "-g" from the Makefile.
This all leads up to the sad thing: gcc-2.0 is /a lot/ bigger than 1.40:
the binary is about 800kB for cc1 alone, and with all the C++ files, I
won't be able to port the whole thing after all, due to lack of disk
space. I gave it one whole 20M partition, and wasn't able to compile
more than the C-part before it ran out of space. I'd guess you'll need
at least 25M free to get everything compiled. That's why I have been
working on the 387-emu: at least whoever ports gcc-2.0 won't have to
trouble with the floating-point patches.
Linus