RE: problem compiling SparseCmplxQR.cc with a RHEL3 box

[Cohen-Tanugi, Johann]

hmmm, looks like my initial response was timed out by this stupid windows 
exchange web server. My apologies if this comes twice :
1) I use this gcc because I used central batch farms which have RHEL3 
installed. True, the admins have installed other compiled versions of gcc
which I can try. gcc 3.2.3 is just the native one. On the other hand, gperf 
2.7.2 is the only one I can find on these machines.
2) I have been browsing the website looking for requirements for compiling from 
source and I found http://wiki.octave.org/wiki.pl?BuildFromSource
which states gcc 3.2+ ..... 
Besides octave 2.9.10 needs only the HEAD of SparseCmplxQR.cc to build with gcc 
3.2.3, and as far as I can tell it is the latest tagged development version....
Now you are telling me that the current HEAD is not gcc 3.2 compatible? Fine, 
but that deserves a mention somewhere, and was that intented, given that the 
latest tag 
is "almost" compatible.... What is the platform on which the CVS directory is 
built? That should also be specified somewhere. I apologized if it is and I am 
just blind.
Looking forward to standing corrected,
Johann



-----Original Message-----
From: John W. Eaton [mailto:address@hidden
Sent: Tue 5/8/2007 6:05 PM
To: Cohen-Tanugi, Johann
Cc: address@hidden
Subject: RE: problem compiling SparseCmplxQR.cc with a RHEL3 box
 
On  8-May-2007, Cohen-Tanugi, Johann wrote:

| good point, sorry! The habit of thinking about linux release rather than 
constituents. So :
| -bash-2.05b$ g++ -v
| Reading specs from /usr/lib/gcc-lib/i386-redhat-linux/3.2.3/specs
| Configured with: ../configure --prefix=/usr --mandir=/usr/share/man 
--infodir=/usr/share/info --enable-shared --enable-threads=posix 
--disable-checking --with-system-zlib --enable-__cxa_atexit 
--host=i386-redhat-linux
| Thread model: posix
| gcc version 3.2.3 20030502 (Red Hat Linux 3.2.3-58)

Why are you still using this version?  It is more than four years old
now and there have been several major releases past 3.2.3.

| but while I was at it I also tried to compile the current HEAD, and I got the 
following compile error :
| oct-gperf.h:55: ISO C++ forbids declaration of `in_word_set' with no type
| oct-gperf.h:93: ISO C++ forbids declaration of `wordlist' with no type
| oct-gperf.h:128: initializer for scalar variable requires one element
| oct-gperf.h:142: ISO C++ forbids declaration of `in_word_set' with no type
| oct-gperf.h: In static member function `static const int*
|    octave_kw_hash::in_word_set(const char*, unsigned int)':
| oct-gperf.h:153: request for member `name' in `wordlist[index]', which is of
|    non-aggregate type `int'
| oct-gperf.h:161: ISO C++ forbids declaration of `wordptr' with no type
| oct-gperf.h:162: ISO C++ forbids declaration of `wordendptr' with no type
| oct-gperf.h:166: request for member `name' in `*wordptr', which is of
|    non-aggregate type `int'
| lex.l: In function `int is_keyword_token(const std::string&)':
| lex.l:998: cannot convert `const int*' to `const octave_kw*' in initialization
| lex.l: In function `octave_value_list Fiskeyword(const octave_value_list&,
|    int)':
| lex.l:2459: request for member `name' in `wordlist[i]', which is of
|    non-aggregate type `int'
| 
| Just passing the info in case it helps, I know what working with the HEAD 
implies.....

I don't think this is a bug in Octave.  Probably these messages are
either because you are using an obsolete version of GCC or because you
also have an obsolete version of gperf.

jwe