A lot of compilers have come and gone over the last years but GCC has been one of the leading compilers in use over the last 15 years. It has a long and critically important history in the free and open source movement. GCC version 4 now features a new optimization framework (Tree-SSA) and includes improvements to its optimizer (e.g. dead code elimination, autovectorisation of loops) as well as language-specific improvements. The changelog (gcc.gnu.org/gcc-4.0/ch...) states: "Independent testers have measured speed-ups up to 25% in real-world production code, compared to the 3.4 family" (for C++). Tree-SSA will enable the development of many more optimizations than were reasonably possible with the old infrastructure - so GCC 4.0 is the base of the next round of optimization, which will be part of GCC 4.1. Tom (www.very-clever.com/)
Re: missing version...
> meanwhile gcc 3.2 is out. why dont they
> update on freshmeat?
meanwhile gcc 3.2 is out. I just made the announcement on fm some minutes ago.
btw. "they" also means "you". :)
meanwhile gcc 3.2 is out. why dont they update on freshmeat?
GCC 3.0.4 got broken stdc++??
Hm... I just did a configure --prefix follow by
make and make install.
Compiled X, KDE and kernel. No problem so far...
Broken libstdc++ in GCC 3.0.4
The version of libstdc++ included with GCC 3.0.4 is somewhat broken.
Specifically, when you attempt to configure it (which normally happens during 'make bootstrap') the configure script complains that you have a broken g++, tells you to upgrade to a version of GCC newer than 3.0, and quits. This leaves you with an effectively broken C++ compiler, since you don't have a C++ library.
The fix is to get a new version of libstdc++. Unfortunately, the GNUheads haven't bothered to put any of the recent versions of libstdc++ (i.e. 3.0.x) on the ftp.gnu.org server, and the gnu web page doesn't refer you to the 'real' project ftp site. So, in order to have a working C++ compiler, get sourceware.cygnus.com/..., and use it to replace the gcc-3.0.4/libstdc++-v3/ directory. Then compile as normal.
Again, I'm sorry about the delay. As for handling different branches of
a project as semi-separate entities, this is in the plans for freshmeat ][,
which scoop is coding as we speak.
We don't make announcements of releases that we can't confirm
The Changelog URL would confirm.
If you had replied to his message and pointed out that the announcement was taking too long to process, I would have checked it myself to see whether I would have better luck.
And I thought that he wouldn't announce it, because announcing a test release after 4-5 doesn't make much sense.
I would recommend changing the Changelog field of the project's appindex record to point to the latest information.
What I did for both announcements. The problem is that Freshmeat is somewhat badly designed. If I add another URL on any field it shows up for any version and also at the appindex, what's wrong. Having different URLs for each announcement and stable/development releases is much better.
I must admit, I was rude (firstname.lastname@example.org is a better place).
Your comment was not deleted. You attached it to the specific
announcement, not to the general appindex page, so it doesn't show up
on freshmeat.net/projects.... I believe comments posted to
specific announcements previously appeared on the page for their
announcements (in this case, freshmeat.net/history/...),
but this doesn't seem to be true at the moment. Here's a copy of your
post for anyone who missed it:
Check here. Consider
this my last contribution to Freshmeat. I posted it 4-5 days ago when
it was released, then someone from Freshmeat replied that he couldn't
connect to the host from his machine (despite both download locations
working without any problems for me on various machines). OK, but 4-5
days to announce it? Give me a break. I really thought that announcing
such test releases would help the GCC team, mainly because this site
is popular, but if it's to get it announced with a huge delay, no
thanks. If it wasn't enough, the announcements are edited and relevant
information is removed, like the URL of the announcement from the
I'm sorry that your submission was delayed. We don't make
announcements of releases that we can't confirm, and I didn't realize
Nathan was still having trouble connecting to the download URLs. If
you had replied to his message and pointed out that the announcement
was taking too long to process, I would have checked it myself to see
whether I would have better luck.
As for the URL in the changes description, we don't allow HTML in our
posts because our data is posted to newsgroups, pulled from
freshmeat.net/backend/, etc. by people who are expecting to
receive plain text. I would recommend changing the Changelog field of
the project's appindex record to point to the latest information.
Censorship? It seems that you can't post a comment with relevant information about an announcement that got edited by a butcher and also say why it was announced 4-5 days after the submission. If you want to delete this comment, also don't forget to remove my account (otherwise I'll post it again), since I won't give a fuck to your site. Have a nice day.
Announcement for 2.95.3.test1
An NFSv2/v3/v4 server running in user space.
A highly productive Java framework for building amazing enterprise Web and mobile applications.