Re: [PATCH] mscms: Use lcms2, when available [try 2]

[Hans Leidekker]

On Sun, 2013-06-02 at 22:49 +0200, Detlef Riekenberg wrote:
> try2:
> import the unchanged icc34.h from the lcms (v1) dev. package.
> (Do we need to add a Wine LGPL licence header for icc34.h?)
>
> The plan is to remove this include, when changing Wine
> to use the new getter/setter functions of lcms2
> (After the release of Wine-1.6)

My plan was to switch to lcms2 soon *after* the release of Wine 1.6. I studied
the API some more and my conclusion is that while it doesn't have all we need
(specifically cmsReadRawTag and cmsWriteRawTag don't do what we want), we can
avoid importing the icc34.h header.

I'm not in favor of adding temporary workarounds. If distributions want to
drop support for lcms1 in 1.6 I'd suggest that they backport the patches.

mscms: Use lcms2, when available [try 2]

[Detlef Riekenberg]

Hans wrote on Monday, 3. June 2013 at 10:30
> On Sun, 2013-06-02 at 22:49 +0200, Detlef Riekenberg wrote:
> > try2:
> > import the unchanged icc34.h from the lcms (v1) dev. package.
> > (Do we need to add a Wine LGPL licence header for icc34.h?)
> >
> > The plan is to remove this include, when changing Wine
> > to use the new getter/setter functions of lcms2
> > (After the release of Wine-1.6)
>
> My plan was to switch to lcms2 soon *after* the release of Wine 1.6.

My plan was to find a way to increase the user experience and provide
lcms2 support out of the box for the stable Wine release.

Alexandre, do you see a chance for lcms2 support for Wine-1.6 (out of the box)?
> I studied the API some more and my conclusion is that while it doesn't have all
> we need (specifically cmsReadRawTag and cmsWriteRawTag don't do what we want),

rc1 for lcms2 v2.5 was just released:
http://www.littlecms.com/lcms2-2.5rc1.tar.gz
What does Wine need from lcms / for cmsReadRawTag and cmsWriteRawTag?
I added [e-mail] for this mail.
> we can avoid importing the icc34.h header.

Hans, you found a way to change Wine to use getter/setter functions
while still linking to lcms v1 and than switch to lcms2?
Great!
Please confirm, that you think, that this is a working transition variant.
(I must admit, that I didn't studied the lcms2 API deep enough).
> I'm not in favor of adding temporary workarounds. If distributions want to
> drop support for lcms1 in 1.6 I'd suggest that they backport the patches.

Adding icc34.h is a variant for a transition to lcms2 without much effort,
while knowing, that this work, but doesn't look beautiful.

Fletching out our mscms header (similar to a stripped down icc34.h),
to allow our current code to compile without icc34.h and link
to lcms2 is another possible transition variant.

Should we spend that time/effort, when we expect, that the added code is
no longer needed after updating Wine to use the lcms2 getter/setter functions?

[Lcms-user] mscms: Use lcms2, when available [try 2]

[Marti Maria]

Hi Guys,

That's Marti Maria, lcms author.

I would be glad to know which functionality you miss in lcms2. I tried 
hard to make
lcms2 complete enough to be a superset of lcms1 but is quite probably I 
left something.

Regarding icc34.h, I would avoid that because licensing issues and more 
importantly
because portability issues. All yous should need is lcms2.h and maybe a 
couple of #defines
for tag names.  Again, please let me know about the individual problems 
you would have.
> (specifically cmsReadRawTag and cmsWriteRawTag don't do what we want),

Ops, so I did something wrong. These functions were in the API to 
implement mscms
and JDK APIs. Could you please elaborate what the issue is?

Thanks again for using lcms.

Best regards
Marti.

El 04/06/2013 0:48, Detlef Riekenberg escribió:
> Hans wrote on Monday, 3. June 2013 at 10:30
> > On Sun, 2013-06-02 at 22:49 +0200, Detlef Riekenberg wrote:
> > > try2:
> > > import the unchanged icc34.h from the lcms (v1) dev. package.
> > > (Do we need to add a Wine LGPL licence header for icc34.h?)
> > >
> > > The plan is to remove this include, when changing Wine
> > > to use the new getter/setter functions of lcms2
> > > (After the release of Wine-1.6)
> >
> > My plan was to switch to lcms2 soon *after* the release of Wine 1.6.
>
> My plan was to find a way to increase the user experience and provide
> lcms2 support out of the box for the stable Wine release.
>
> Alexandre, do you see a chance for lcms2 support for Wine-1.6 (out of the box)?
> > I studied the API some more and my conclusion is that while it doesn't have all
> > we need (specifically cmsReadRawTag and cmsWriteRawTag don't do what we want),
>
> rc1 for lcms2 v2.5 was just released:
> http://www.littlecms.com/lcms2-2.5rc1.tar.gz
> What does Wine need from lcms / for cmsReadRawTag and cmsWriteRawTag?
> I added [e-mail] for this mail.
> > we can avoid importing the icc34.h header.
>
> Hans, you found a way to change Wine to use getter/setter functions
> while still linking to lcms v1 and than switch to lcms2?
> Great!
> Please confirm, that you think, that this is a working transition variant.
> (I must admit, that I didn't studied the lcms2 API deep enough).
> > I'm not in favor of adding temporary workarounds. If distributions want to
> > drop support for lcms1 in 1.6 I'd suggest that they backport the patches.
>
> Adding icc34.h is a variant for a transition to lcms2 without much effort,
> while knowing, that this work, but doesn't look beautiful.
>
> Fletching out our mscms header (similar to a stripped down icc34.h),
> to allow our current code to compile without icc34.h and link
> to lcms2 is another possible transition variant.
>
> Should we spend that time/effort, when we expect, that the added code is
> no longer needed after updating Wine to use the lcms2 getter/setter functions?

[Lcms-user] mscms: Use lcms2, when available [try 2]

[Hans Leidekker]

On Tue, 2013-06-04 at 01:36 +0200, Marti Maria wrote:
> > (specifically cmsReadRawTag and cmsWriteRawTag don't do what we want),
>
> Ops, so I did something wrong. These functions were in the API to 
> implement mscms
> and JDK APIs. Could you please elaborate what the issue is?

They don't allow reading or writing to an offset in the tag data. We could
of course work around that by buffering all of the data, but in this case
I think it's cleaner to stick with our own implementation, ported to the
lcms2 header.

Thank you for your efforts to support Wine better, lcms2 is a real improvement.

mscms: Use lcms2, when available [try 2]

[Hans Leidekker]

On Tue, 2013-06-04 at 00:48 +0200, Detlef Riekenberg wrote:
> Hans, you found a way to change Wine to use getter/setter functions
> while still linking to lcms v1 and than switch to lcms2?

No, I'm not interested in doing that.