Re: [elinks-dev] [PATCH] bug 764: Fix option_value initialization on sparc64

2009-08-10 Thread Kalle Olavi Niemitalo
Edd Barrett  writes:

> Do you have a patch?

Please test the 0.11/bug764 branch shown here:
http://repo.or.cz/w/elinks/kon.git
That should contain the required commits from elinks-0.12 and a
simple patch that fixes union option_value when ELinks start up.
(It makes some assumptions outside of standard C, but at least
the code should be more portable than in ELinks 0.11.6.)

I may make some further changes to the branch if I find a bug
there.  (That seems unlikely to happen before Friday though.)
When reporting results, please mention the commit ID you tested.

> I could try it on OpenBSD, which is mostly POSIX compatible.

If OpenBSD has dlsym(), then casts between data pointers and
function pointers have to work.

> Just out of interest, how stable is elinks "unstable".

For the last 12 months, there have been plenty of bug fixes
and almost no new features in elinks-0.12.  Crashes are rare,
especially if you keep ECMAScript disabled as is now the default.
I've been trying to keep track of regressions compared to
elinks-0.11; only bug 1066 (in IDN) is left, and I suspect
elinks-0.11 has a similar bug itself.  However, the future fixes
for bugs 1066 and 771 may cause new bugs.


pgpTLth9IG970.pgp
Description: PGP signature
___
elinks-dev mailing list
[email protected]
http://linuxfromscratch.org/mailman/listinfo/elinks-dev


Re: [elinks-dev] [PATCH] bug 764: Fix option_value initialization on sparc64

2009-08-09 Thread Edd Barrett
On Sun, Aug 09, 2009 at 07:15:09PM +0300, Kalle Olavi Niemitalo wrote:
> Edd Barrett  writes:
> 
> > My next question is, how hard is it to put this back into the stable
> > branch?
> 
> This patch itself wouldn't be too hard to apply to elinks-0.11.

I don't think I have the time to do so. Will you be planning to do this
anyway? I think it would be wise, given that *all* sparc64 machines
independent of platform will be unusable.


> A much simpler patch is also possible if we can assume that
> casting function pointers or integers to data pointers and back
> preserves the values.  (POSIX guarantees this for function
> pointers but ELinks has been run in non-POSIX systems in the
> past.  AFAIK, nothing guarantees this for data pointers.)

Is there such a thing as a POSIX compatible system? :P

Do you have a patch? I could try it on OpenBSD, which is mostly POSIX
compatible.

> 
> However, some other parts of bug 764 have already been fixed in
> elinks-0.12.  So we'd first need to make a list of those commits and
> backport them.  Commit 1ec1dc43be13e062e18368bf05928ac95d7b7ba5 is
> one of them for sure.

Just out of interest, how stable is elinks "unstable". As a rule of
thumb OpenBSD would not accept "unstable" code, but maybe if backporting
will be painful, it could be possible to use a snapshot for our packages
system.

Thanks

-- 

Best Regards

Edd Barrett
(Freelance software developer / technical writer / open-source developer)

http://students.dec.bmth.ac.uk/ebarrett
___
elinks-dev mailing list
[email protected]
http://linuxfromscratch.org/mailman/listinfo/elinks-dev


Re: [elinks-dev] [PATCH] bug 764: Fix option_value initialization on sparc64

2009-08-09 Thread Kalle Olavi Niemitalo
Edd Barrett  writes:

> My next question is, how hard is it to put this back into the stable
> branch?

This patch itself wouldn't be too hard to apply to elinks-0.11.
A much simpler patch is also possible if we can assume that
casting function pointers or integers to data pointers and back
preserves the values.  (POSIX guarantees this for function
pointers but ELinks has been run in non-POSIX systems in the
past.  AFAIK, nothing guarantees this for data pointers.)

However, some other parts of bug 764 have already been fixed in
elinks-0.12.  So we'd first need to make a list of those commits and
backport them.  Commit 1ec1dc43be13e062e18368bf05928ac95d7b7ba5 is
one of them for sure.


pgpEqehM1AR3v.pgp
Description: PGP signature
___
elinks-dev mailing list
[email protected]
http://linuxfromscratch.org/mailman/listinfo/elinks-dev


Re: [elinks-dev] [PATCH] bug 764: Fix option_value initialization on sparc64

2009-08-09 Thread Edd Barrett
On Sun, Aug 09, 2009 at 12:20:17AM +0300, Kalle Olavi Niemitalo wrote:
> Kalle Olavi Niemitalo  writes:
> 
> > Please try this patch.

This fixes all of the issues I reported. :)

My next question is, how hard is it to put this back into the stable
branch?

-- 

Best Regards

Edd Barrett
(Freelance software developer / technical writer / open-source developer)

http://students.dec.bmth.ac.uk/ebarrett
___
elinks-dev mailing list
[email protected]
http://linuxfromscratch.org/mailman/listinfo/elinks-dev


Re: [elinks-dev] [PATCH] bug 764: Fix option_value initialization on sparc64

2009-08-08 Thread Kalle Olavi Niemitalo
Kalle Olavi Niemitalo  writes:

> Please try this patch.

The patch is for elinks-0.12.  I put it in Bugzilla as well:
http://bugzilla.elinks.cz/attachment.cgi?id=547&action=view


pgpe7dcHDoGaJ.pgp
Description: PGP signature
___
elinks-dev mailing list
[email protected]
http://linuxfromscratch.org/mailman/listinfo/elinks-dev