This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [Proposed binutils PATCH] Re: Diagnosing an intricate C++ problem


In article <20000905175437.H10417@Jeffreys.suse.de> you write:
>* H . J . Lu (hjl@lucon.org) [20000905 17:45]:

>> Why not xxxBSD just spend some effort to make sure the main trunck is
>> working for you?

>You seem to completely ignore the way *BSD trees are built/updated. There
>are some people that object to tracking moving targets, i.e. CVS versions.
>Not all the world is Linux.

>And IMHO even a good part of the Linux world would appreciate having only
>one source for binutils.

Speaking from the OpenBSD point of view:

From my point of view, binutils has indeed been a moving target between
binutils-2.9.1 and binutils-2.10.

A few years without a release is a looong time.

We are willing to do some testing, but using snapshots from the development
tree is not ok, specifically because there are bandwidth issues with some
developers, and communication issues: getting them to work with a cvs
snapshot is hard, especially since not everyone has permanent high
bandwidth network connection.

Integrating a snapshot on the actual OpenBSD cvs repository is hard as well,
especially since we have sound rules that specify that what's in the tree
should NOT break, and should work on all platforms we support.

In short, having the latest, actual, release, be something that is pretty 
ancient  is a large problem for us.

H.J.Lu is probably spending quite a large amount of time on linux binutils.
We don't have one *single* developer who can match that amount of time.
But, if we put all our forces together, we probably come pretty close.

Of course, this means a much larger level of internal coordination.


Also, H.J., *please* refrain to use the encompassing xxxBSD term.
There are common points between all BSD projects, but there are also
fairly large differences.  You might have had experiences in the past
with BSD people... well, guess what ? a large number of new people
exist now. OpenBSD is not that old. We forked from NetBSD a few years
back (I wasn't on board then), and we've been running ever since.
We *still* have a few legacy issues that go back to NetBSD, especially
some binutils code that has never been folded back to the FSF...
this is rather painful, as nothing short of a complete rewrite of that
code can save us (I've asked NetBSD people on this. The answers I've got
were either inexistant, or fairly vague, and I never got the kind of 
follow-up I would have wanted).

Hopefully, we should finally be able to switch to an almost `completely FSF'
release of binutils soon.

The main reason this can finally happen is because binutils 2.10 came out.
There had been some significant effort to clean the slate before, but there
are various logistical reasons why this could not be achieved with a 
binutils snapshot.

So, I, for one, am hoping that binutils releases are going to happen at a
somewhat faster pace... 

... and I'm also quite willing to cooperate with the FSF on binutils 
development. Right now, my efforts have been twarted by this all-powerful
force: the FSF bureaucracy (my copyright assignments for binutils and 
gdb still have not been recorded, makes sense, since I sent them only
one month ago, silly me)

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]