[Date Prev][Date Next][Thread Prev][Thread Next][Thread Index]

RE: [XaraXtreme-dev] Problem with contours loading Groucho2.xar



Err... That doesn't make sense. The addition of the pentium4 flag was my
mistake. I hadn't meant to check in that change. Hence Luke removed the
flag. So older versions of CDraw didn't use the flag either, but these
worked. It sounds to me like some files haven't been rebuilt after the
removal of the flag, but I suspect that is unlikely.

The only code that relied on the flag was code that has now been removed
and has nothing to do with path contouring anyway.

Gavin

-----Original Message-----
From: owner-dev@xxxxxxxxxxxxxxxx [mailto:owner-dev@xxxxxxxxxxxxxxxx] On
Behalf Of Gerry Iles
Sent: 27 June 2006 16:45
To: dev@xxxxxxxxxxxxxx
Subject: RE: [XaraXtreme-dev] Problem with contours loading Groucho2.xar

Well, it turns out that r1354 works and r1355 doesn't.  r1355 was just a
new build of the CDraw library that didn't use the march=pentium4 flag
but was otherwise built from the same source as the version added in
r1342.

It would appear that there is something in the r61 CDraw update that
doesn't work correctly on x86 unless march=pentium4 is specified.

This does explain why x86_64 builds don't show the same problem.

Gerry

-----Original Message-----
From: owner-dev@xxxxxxxxxxxxxxxx [mailto:owner-dev@xxxxxxxxxxxxxxxx] On
Behalf Of Alex Bligh
Sent: 27 June 2006 13:24
To: dev@xxxxxxxxxxxxxx
Cc: Alex Bligh
Subject: RE: [XaraXtreme-dev] Problem with contours loading Groucho2.xar



--On 27 June 2006 13:01 +0100 Phil Martin <Phil@xxxxxxxx> wrote:

> This sounds like exactly the same bug as BZ1192 which also only
appeared
> recently and whose sample doc worked when loaded into a version of LX
> that had RGB/BGR issues (not sure which version).
>
> I haven't got very far with it yet, except to ask Gavin and Gavin
> suggests LX is passing CDraw bad data.

It does sound like the same issue: I can replicate BZ1192 on i386 debug
and not x86_64 debug, just like Gerry's problem.

I can't see it's likely to be /caused/ by RGB/BGR swap though - just
a contemporaneous change.

Alex