[webkit-dev] Announcing new port: Nix

Osztrogonác Csaba oszi at inf.u-szeged.hu
Thu Sep 12 03:53:11 PDT 2013


Hi,

Geoffrey Garen írta:
>>>        Does your code fix bugs or improve performance in WebCore or 
>>> JavaScriptCore?
>> As any other WebKit port WebCore bugs are fixed when we found it and
>> need a fix, our current performance focus is on CoordinatedGraphics
>> and there are another team working on get the canvas faster.
> 
> I don't share your premise that all ports, by their nature, improve the 
> correctness and performance of core code.
> 
> Can you point to some important correctness and/or performance 
> improvements WebKit will see if it adopts your code into trunk?

[snip]

> In that case, what is the benefit to WebKit?
> 
> Geoff

Sure. We at University of Szeged who work on WebKitNix from this year
contributed many fixes to various part of WebKit: JavaScriptCore,
WebCore, Tools, ... We landed all the fixes to WebKit trunk whenever
possible. Let me look for some fixes without striving for completeness:

Improving cURL network backend (which used by Apple's WinCairo port too)
-------------------------------------------------------------------------
- https://trac.webkit.org/changeset/150118
- https://trac.webkit.org/changeset/150221
- https://trac.webkit.org/changeset/150964
- https://trac.webkit.org/changeset/151195
- https://trac.webkit.org/changeset/151238
- https://trac.webkit.org/changeset/151254
- https://trac.webkit.org/changeset/151255
- https://trac.webkit.org/changeset/151346
- https://trac.webkit.org/changeset/151348
- https://trac.webkit.org/changeset/151441
- https://trac.webkit.org/changeset/151580
- https://trac.webkit.org/changeset/151668

JavaScriptCore fixes
---------------------
- https://trac.webkit.org/changeset/142481
- https://trac.webkit.org/changeset/142616
- https://trac.webkit.org/changeset/143346
- https://trac.webkit.org/changeset/145047
- https://trac.webkit.org/changeset/145194
- https://trac.webkit.org/changeset/145505
- https://trac.webkit.org/changeset/145551
- https://trac.webkit.org/changeset/145689
- https://trac.webkit.org/changeset/146309
- https://trac.webkit.org/changeset/146396

WebInspector fixes
-------------------
- https://trac.webkit.org/changeset/153471
- https://trac.webkit.org/changeset/153044
- https://trac.webkit.org/changeset/151217
- https://trac.webkit.org/changeset/150578
- https://trac.webkit.org/changeset/150959

Various assertion/missing null check fixes
-------------------------------------------
- https://trac.webkit.org/changeset/146083
- https://trac.webkit.org/changeset/153433
- https://trac.webkit.org/changeset/153883
- https://trac.webkit.org/changeset/154439
- https://trac.webkit.org/changeset/154476
- https://trac.webkit.org/changeset/154698

I think these changes/fixes are benefit to the WebKit project. Some of
them for all ports, some of them for only Linux users, some of them for
only ARM devices, ...

Reni's Fuzzinator, a mutation based web fuzzer already revealed 75 new
bugs. (28 of them are already fixed, there are 47 opened now)
See https://bugs.webkit.org/show_bug.cgi?id=116980 for details.

Additionally we landed at least dozens, but maybe hundreds buildfixes
after accidental, forgetful (and intentional) breakages too. Not to
mention the hundreds bug reports about new regressions/failures and
how many times we helped other developers with backtraces, trying
WIP fixes on platforms they don't have.

I think the INdT guys could compose a long list of their general
fixes landed in WebKit trunk too if it is necessary.

Ossy


More information about the webkit-dev mailing list