-Document Really Ugly Bugs (coredumps, assert fails, etc)
+This documents serious bugs
-[GNU libc]
+********
-The GNU extension memmem() is known to be buggy on linux libc 5.0.9
-and before. Glibc upto 2.0.5 also has problems with memmem (), but
-these should not affect LilyPond.
+[LinuxPPC-R5, egcs-1.1.2-12c]
+Serious egcs-1.1.2-12c (stock LinuxPPC R5) bug on ppc:
-[IRIX (5.3?)]
+ *dest++ = *src++;
-coredump from strstream::strstream () upon the first read of a file
+Bug report filed, fixed in egcs-1.1.2-12f.
+Although we've currently got a workaround in place, the compiler
+is buggy; you should upgrade:
-[Linux Intel]
+ ftp://dev.linuxppc.org/users/fsirl/R5/RPMS/ppc/
+
+
+[LinuxPPC-R4, egcs-1.0.2]
+
+All compiling with -O2 is suspect, in particular guile-1.3, and
+Lily herself will break.
+
+
+[LinuxPPC, egcs-1.1b]
+
+Lilypond 1.1.35-1.1.51,...: Wierd errors in guile scm_gc_mark ()
+
+
+[All platforms]
+
+Some bugs may be captured in input/bugs/*y
+
+[Linux i386, RedHat 5.2 with updates to 6.0]
+
+Compiling with
+
+ configure --disable-checking --enable-printing --disable-optimise --disable-debugging
+
+results in core dumps, during parsing of init files. Cause unknown.
+Solution: use
+
+ --enable-checking and --enable-optimize
+
+[Linux libg++ 2.7]
LilyPond occasionally crashes while parsing the initialisation files.
This is a very obscure bug, and usually entering the commandline
Includable_lexer::new_input (this=0x8209a00, s={strh_ = {
:
+This behaviour has been observed with machines that have old libg++
+versions (LinuxPPC feb '98, RedHat 4.x).
-I get bitten by this every once in a while, and I am very interested
-in hints what might be wrong. This problem has only been identified
-with libc-5.3 and libc-5.4 platforms, so you might try upgrading to
-6.0, ie. GNU libc-2.
[Linux Intel]