On Sat, 2016-10-15 at 22:58 +0300, clutton wrote:
> On Fri, 2016-10-14 at 06:42 +0300, clutton wrote:

Just fixed one major issue. Those user observers left, it would compile
 and work if two strings was commented but I would prefer to make
appropriate patch, without cutting unknown.


Could we please stop making new commits to the tree till this is
merged? What is the point patching the old version if new is coming?

When this would be ready, it would be very advisable to clean old
patches. Some things are already in decent state in chromium codebase,
no need to hack it around.

Some approaches which I followed and think would be good:
1) GN: is_bsd, not is_freebsd, is_somebsd, just use one identifier.
2) C++ files: _bsd.cc, not _freebsd.cc and _openbsd.cc, the
implementation can be divided inside like they divide
Linux/Android/ChromeOS in one *.cc file. But we are far away from
3) C++ code: generally OS_BSD, in some cases OS_SOMEBSD when it is
really needed.
4) Python: well... just ugly and I have no idea what to do with this
now. A lot of different calls and approaches already exist in code. But
fortunately it's not very important now and everyone could write

P.S. Pathes are mostly clean (I know where they aren't and would clean
them when major issues are resolved), the Makefile... just use it for
now it wouldn't be hard to make it clean...

P.P.S. Contact me if you have any questions.

P.P.P.S. Advantages of new build: GN works really fast, generation
ninja files takes much less time. Linking with lld is extremely fast
and doesn't require a lot of RAM.
Disadvantages: currently it uses almost all libraries from third_party.

Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to