Questions about the font shaping tender

Khaled Hosny khaledhosny at
Tue Feb 2 11:58:55 CET 2016

On Tue, Feb 02, 2016 at 10:15:07AM +0100, Miklos Vajna wrote:
> Hi,
> I think the problem to solve is that we have different layout on Linux,
> Windows and OS X, which means it's next to impossible to write unit
> tests that assert the text is laid out correctly, since there can be
> small differences today. The hope is that this is the root cause of e.g.
> the chart tests which are not enabled by default today.
> As far as I know, there are two ways you can solve this:
> - use the Linux stack everywhere (fontconfig, freetype, harfbuzz?
>   something like that), IIRC that's what e.g. Chromium does
> - create a 4th layout in VCL, which is just a stub, but gives exactly
>   the same results in all platforms, to allow writing unit tests that
>   are executed in all platforms

OK, I think I see the line of reasoning behind the second option, but I
still fail to understand what AFM has to do with this, using AFM to get
glyph metrics to avoid using platform specific APIs? If it is just a
stub implementation, why not use a fixed width font and simplify/dump
things even further?

Personally I’d do a mixture of the two, a 7th (8th? I think we have 6 or
7 SalLayout implementations already) SalLayout that uses HarfBuzz
(reusing the existing HarfBuzz integration code as much as possible)
then gradually switch existing platforms to it as it matures, so it is
an overall win not just for the sake of fixing tests.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: not available
URL: <>

More information about the LibreOffice mailing list