egcs-1.0 Release

egcs-1.0 Release

[Jeffrey A Law]

We are pleased to announce the release of egcs-1.0. 

egcs is a collaborative effort involving several groups of hackers using an
open development model to accelerate development and testing of GNU compilers
and runtime libraries. 

An important goal of egcs is to allow wide scale testing of experimental
features and optimizations; therefore, egcs contains some features and
optimizations which are still under development. However, egcs has been
carefully tested and should be comparable in quality to most gcc releases. 

egcs-1.0 contains many improvements and features not found in gcc-2.7 and
even the soon to be released gcc-2.8 compilers. 

   * Integrated C++ runtime libraries, including support for most major
     linux systems! 
   * The integrated libstdc++ library includes a verbatim copy of
     SGI's STL release. 
   * Integrated GNU Fortran compiler 
   * New instruction scheduler 
   * New alias analysis code 

See the egcs web page for a more complete list of new features, installation
instructions, build/test status, caveats, mailing lists, etc.
http://www.cygnus.com/egcs

You can download egcs-1.0 directly from
ftp.cygnus.com:/pub/egcs/releases/egcs-1.0

The release should be available on the egcs mirrors sites shortly.
See the egcs home page for a list of mirror sites.

We'd like to thank the numerous people that have contributed new features,
test results, bugfixes, documentation updates, web page updates, general
suggestions, etc.  Unfortunately, they're far too numerous to mention by
name.


Questions, comments, etc should be directed to egcs@cygnus.com
Bug reports should be directed to egcs-bugs@cygnus.com

Re: egcs-1.0 Release

[Tudor Hulubei]

Hi,

I just tried the new snapshot (971207) and it seems that the memory
problem when using lots of templates is still present.  cc1plus goes
up to about 138Mb and then sometimes reports "memory exhausted",
sometimes dies with an "internal compiler error".

Regards,
Tudor

memory explosion with -O1 (was: egcs-1.0 Release)

[Gerald Pfeifer]

On Mon, 8 Dec 1997, Tudor Hulubei wrote:
> I just tried the new snapshot (971207) and it seems that the memory
> problem when using lots of templates is still present.  cc1plus goes
> up to about 138Mb and then sometimes reports "memory exhausted",

I can confirm this report:

On sparc-sun-solaris2.5.1 with egcs-2.91.02 971206 I experience exactly
the same problem with "-O1 -ftemplate-depth-20" only, no -Wall etc!

I have reported this problem, which doesn't occur with either 
gcc 2.7.2 or egcs without optimizations, before. (It's available 
as http://www.cygnus.com/ml/egcs-bugs/1997-Nov/0142.html .)

If anyone would like to have a look at my sources, just let me know!

Gerald
-- 
Gerald Pfeifer (Jerry)      Vienna University of Technology
pfeifer@dbai.tuwien.ac.at   http://www.dbai.tuwien.ac.at/~pfeifer/

memory explosion with -O1 (was: egcs-1.0 Release)

[Mark Mitchell]

Gerald --

>>>>> "Gerald" == Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> writes:

    Gerald> On Mon, 8 Dec 1997, Tudor Hulubei wrote:
    >> I just tried the new snapshot (971207) and it seems that the
    >> memory problem when using lots of templates is still present.
    >> cc1plus goes up to about 138Mb and then sometimes reports
    >> "memory exhausted",

    Gerald> I can confirm this report:

    Gerald> On sparc-sun-solaris2.5.1 with egcs-2.91.02 971206 I
    Gerald> experience exactly the same problem with "-O1
    Gerald> -ftemplate-depth-20" only, no -Wall etc!

    Gerald> I have reported this problem, which doesn't occur with
    Gerald> either gcc 2.7.2 or egcs without optimizations,
    Gerald> before. (It's available as
    Gerald> http://www.cygnus.com/ml/egcs-bugs/1997-Nov/0142.html .)

    Gerald> If anyone would like to have a look at my sources, just
    Gerald> let me know!

Please send me, and just me, not the list, the .ii file generated when
you compile with -save-temps.  I'll see if I can figure anything out
about where the memory is going.  I don't guarantee anything, but I'll
see what I can do.

Again, please be sure not to send the preprocessed source to this
list, since you earlier message indicated that it was quite large.

Thanks,

-- 
Mark Mitchell		mmitchell@usa.net
Stanford University	http://www.stanford.edu