Re: sfront 0.56 update ...

From: John Lazzaro (lazzaro@cs.berkeley.edu)
Date: Tue Mar 07 2000 - 19:01:40 EST


> I am now a little puzzled, as I have had one problem, but not this one!
> I have built sfront56 (release) from the winlite sources, and (a) it
> builds, and (b) it compiles 'beat' into sa.c which compiles and runs,
> and generates a nice soundfile. I am using VC++ v5.0 with Service Pack 3
> (without which all hell breaks loose...).

Thanks so much for this report, it helps narrow things
down a lot -- basically:

-- Daniel Kahlin reports that the binary shipped with
sfrontwin55 (created with dos-gcc, a DOS cross-compiler
based on gcc 2.7.2.1) also has this problem, so its not
new with 0.56

-- I tried compiling 0.56 on WinNT using a gcc of similar
vintage (2.7.X) and found the same problem, so its not
dos-gcc specific.

-- Kees van Prooijen reported a happy 0.55 sfront build
using Borland IDE, another datapoint showing 0.55 is
basically OK on Windows with a non gcc-derived compiler.

-- All the UNIX tests are fine -- multiple compilers on
multiple platforms -- on 0.56.

-- My tests today on NT are pointing towards a blowup
in code bison generates, during a malloc ...

Given all this, I'd say Borland and MS C++ Win users
are probably pretty safe in just downloading the
"generic source" sfront 0.56 on the website, compiling
it with the commercial compiler, and tune out of this
discussion :-). For the folks using gcc, the plan of
attack is as follows:

-- Try 0.54 and see if that works:

http://www.cs.berkeley.edu/~lazzaro/sa/archive/sfront0.54win.zip

That's the last sfront with a pre-built windows binary, if that
one still doesn't work increment backwards using:

http://www.cs.berkeley.edu/~lazzaro/sa/archive/sfront0.53lite.tar.gz
http://www.cs.berkeley.edu/~lazzaro/sa/archive/sfront0.52lite.tar.gz

[...]

http://www.cs.berkeley.edu/~lazzaro/sa/archive/sfront0.44lite.tar.gz

Since 0.44 came out on Oct 31, I'd be pretty surprised if it had the
bug also! If anyone does do this, please send me an email and let
me know the first good one, it will be quite useful for debugging!

                                                        --jl



This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 11:46:37 EST