freetype-devel
[Top][All Lists]
Advanced

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

Re: [ft-devel] freetype and harfbuzz


From: Vincent Torri
Subject: Re: [ft-devel] freetype and harfbuzz
Date: Mon, 14 Nov 2016 18:09:41 +0100

On Mon, Nov 14, 2016 at 2:09 PM, suzuki toshiya
<address@hidden> wrote:
> Werner LEMBERG wrote:
>> The main question is: Is it *really* worth the hassle?
>
> I was thinking there are several existing software supporting
> "built-but-not-installed-yet" tree by pkg-config, so it would
> not be so difficult for Unix-like systems. But...
>
>>> Actually, it is exactly the case a friend has, and I am asking this
>>> question for him because he is a bit busy.  He is working on an
>>> installer for Windows.  But it seems that the design of his
>>> installer makes difficult to resolve the circular dep with the 3
>>> steps mentioned above.
>
> Oh, the target is Windows! I cannot help much. I don't know how
> Windows developer builds harfbuzz (except of the cases with cygwin
> & mingw).

cross compilation on Linux using mingw-w64, targetting 32 and 64 bits

>> This would be an excellent opportunity to improve his installer :-)
>>
>> Note that it isn't a real circular dependency, since FreeType works
>> just fine without HarfBuzz; it just has limited auto-hinter module.
>>
>> I suggest that you look around how other building systems (like
>> Homebrew for OS X) are resolving the issue.  IIRC, some provide two
>> FreeType packages with different build flags:
>>
>>   freetype-simple -> harfbuzz
>>   harfbuzz -> freetype-complete
>>
>> where freetype-complete simply overwrites the freetype-simple library
>> after installation.
>
> Indeed. If the target is Windows, there would not be system-wide
> harfbuzz nor freetype, so building temporal freetype-simple would
> not be so harmful action (although I accept troublesome works).

i've told my friend about that. I think that it is possible as he is
building the toolchain (binutils, gcc, etc...)

>>> so if Freetype2 supports the extraction of harfbuzz source code and
>>> build it, it would indeed solve his problem
>>
>> Basically, I don't object if this gets done in a clean way, this is,
>> without increasing the FreeType library size and having manageable
>> code that doesn't need to be updated constantly if HarfBuzz evolves.
>> However, I won't be able to work on this (and my interest is quite
>> limited, admittedly).  Any volunteers welcomed!
>
> I don't think duplicating the codes from harfbuzz to FreeType2 is
> good idea, because it is written in C++, and it's development is
> unfinished yet (as written).

ok

thank you

Vincent Torri


> Regards,
> mpsuzuki
>



reply via email to

[Prev in Thread] Current Thread [Next in Thread]