June 2008 - webkit-dev - lists.webkit.org

Qt's default webkit user agent
by Benjamin Meyer 01 Dec '08

01 Dec '08

As it is today Qt's default user agent (in qwebpage.cpp) is hard coded to: "Mozilla/5.0 (Macintosh; U; Intel Mac OS X; en) AppleWebKit/523.15 (KHTML, like Gecko) Safari/419.3 Qt" Seeing as this library will be used inside of many other applications it makes sense to include by default at least the application name/version when they are set. QCoreApplication gives us several strings which can be used. applicationName, applicationVersion, organizationDomain, organizationName But at a bigger question, what should be included (or not included) with the default user agent? Is there docs anywhere that suggest what webkit library users should use as their user agent string? -Benjamin Meyer

7 6

proxying thru curl
by Giri Rao 01 Dec '08

01 Dec '08

Hello, I have set up curl to go through a simple http proxy that I have written. I have a .curlrc with the following line in it: --proxy=http://<myhost>:<myport>/ When I invoke curl on command line, it goes thru the set proxy. However, when I launch webkit through GtkLauncher, it does not seem to go through the proxy at all. Is my understanding wrong in that webkit uses curl to fetch content? How do I make webkit go through my proxy? Any help is appreciated. Thanks Giri

2 2

Re: [webkit-changes] [24723] trunk/WebCore
by Darin Adler 11 Sep '08

11 Sep '08

I'm not happy with the code organization here. XMLTokenizer now has tons of ifdefs and two separate implementations. It's fine to have a QXmlStream implementation, but the two implementations should be in separate files, side by side, as we do in the platform layer. Or the XML parser should be abstracted out so we have a single XMLTokenizer built on a single parser abstraction. I think this is also the type of patch that should be reviewed by at least one person other than the folks concentrating on the Qt port. With the huge number of check-ins today, I'm sure there are other interesting things like this that I missed. -- Darin

8 17

renaming ASSERT macro
by Jörg Bornemann 02 Jul '08

02 Jul '08

Hi *, I'd like to rename the ASSERT macro from JavaScriptCore/wtf/Assertions.h to something that doesn't clash with the Win32 ASSERT macro. There's already a hack that undefines this macro before the definition in Assertion.h. This is an easy work-around but on Windows CE (where I'm developing on) the situation is worse: ASSERT gets defined in windows.h That means ATM that I have to make sure to always include <windows.h> _before_ Assertion.h. Otherwise I come up with a mix of Windows assertions and WebKit assertions, not to speak of the warning of an already defined ASSERT... As you can see in Assertions.h, renaming ASSERT is not completely my idea: /* FIXME: Change to use something other than ASSERT to avoid this conflict with win32. */ So, what do you think about this? What would be a proper name for the macro? WKASSERT? Best Regards, Joerg

5 11

Building Webkit with cygwin failed
by tyl 01 Jul '08

01 Jul '08

Hi all, I'm trying to build webkit on windows XP using cygwin. I used Visual studio 2005 and also installed service pack(VS80sp1-KB926601-X86-ENU.exe), and I was build in debug mode. I followed the instruction on the webkit homepage but when i build i keep getting this error when building webkit: Please help me. here is the FindSafari's build log: Build Log Build started: Project: FindSafari, Configuration: Release|Win32 Command Lines Build machine environment variables may not be correct. Build machine environment variables may not be correct. Creating temporary file "C:\cygwin\home\teng\WebKit\WebKitBuild\obj\FindSafari\Release\RSP00000937282920.rsp" with contents [ /O2 /D "NDEBUG" /D "WIN32" /D "_WINDOWS" /D "_WIN32_WINNT=0x500" /D "_CRT_SECURE_NO_DEPRECATE" /D "_SCL_SECURE_NO_DEPRECATE" /D "__PRODUCTION__=0" /D "_HAS_EXCEPTIONS=0" /D "_UNICODE" /D "UNICODE" /FD /MD /GR- /Fo"C:\cygwin\home\teng\WebKit\WebKitBuild\obj\FindSafari\Release\\" /Fd"C:\cygwin\home\teng\WebKit\WebKitBuild\obj\FindSafari\Release\vc80.pdb" /W4 /WX /c /Zi /TP /wd4018 /wd4068 /wd4099 /wd4100 /wd4127 /wd4138 /wd4189 /wd4244 /wd4275 /wd4291 /wd4344 /wd4355 /wd4389 /wd4503 /wd4510 /wd4512 /wd4610 /wd4706 /wd4800 /wd4951 /wd4952 /wd4996 /wd6011 /wd6031 /wd6211 /wd6246 /wd6255 /wd6387 /GS .\FindSafari.cpp ] Creating command line "cl.exe @C:\cygwin\home\teng\WebKit\WebKitBuild\obj\FindSafari\Release\RSP00000937282920.rsp /nologo /errorReport:queue" Creating temporary file "C:\cygwin\home\teng\WebKit\WebKitBuild\obj\FindSafari\Release\BAT00000A37282920.bat" with contents [ @echo off if not exist "C:\cygwin\home\teng\WebKit\WebKitBuild\public\sym" mkdir "C:\cygwin\home\teng\WebKit\WebKitBuild\public\sym" if errorlevel 1 goto VCReportError goto VCEnd :VCReportError echo Project : error PRJ0019: A tool returned an error code from "Performing Pre-Link Event..." exit 1 :VCEnd ] Creating command line "C:\cygwin\home\teng\WebKit\WebKitBuild\obj\FindSafari\Release\BAT00000A37282920.bat" Creating temporary file "C:\cygwin\home\teng\WebKit\WebKitBuild\obj\FindSafari\Release\RSP00000B37282920.rsp" with contents [ /OUT:"C:\cygwin\home\teng\WebKit\WebKitBuild\bin\FindSafari.exe" /INCREMENTAL:NO /LIBPATH:"C:\cygwin\home\teng\WebKit\WebKitBuild\lib" /LIBPATH:"C:\cygwin\home\teng\WebKit\WebKitLibraries\win\lib" /MANIFEST /MANIFESTFILE:"C:\cygwin\home\teng\WebKit\WebKitBuild\obj\FindSafari\Release\FindSafari.exe.intermediate.manifest" /DEBUG /PDB:"c:\cygwin\home\teng\webkit\webkitbuild\bin\FindSafari.pdb" /PDBSTRIPPED:"C:\cygwin\home\teng\WebKit\WebKitBuild\public\sym\FindSafari.pdb" /SUBSYSTEM:CONSOLE /OPT:REF /OPT:ICF /OPT:NOWIN98 /IMPLIB:"C:\cygwin\home\teng\WebKit\WebKitBuild\lib\FindSafari.lib" /MACHINE:X86 /SAFESEH /FIXED:NO /dynamicbase shlwapi.lib advapi32.lib ole32.lib kernel32.lib user32.lib gdi32.lib winspool.lib comdlg32.lib advapi32.lib shell32.lib ole32.lib oleaut32.lib uuid.lib odbc32.lib odbccp32.lib "..\..\webkitbuild\obj\findsafari\release\FindSafari.obj" ] Creating command line "link.exe @C:\cygwin\home\teng\WebKit\WebKitBuild\obj\FindSafari\Release\RSP00000B37282920.rsp /NOLOGO /ERRORREPORT:QUEUE" Output Window Compiling... FindSafari.cpp Performing Pre-Link Event... Linking... link: missing operand after `ÿþ/' Try `link --help' for more information. Project : error PRJ0002 : Error result 1 returned from 'C:\cygwin\bin\link.exe'. Project : warning PRJ0018 : The following environment variables were not found: $(PRODUCTION) Results Build log was saved at "file://C:\cygwin\home\teng\WebKit\WebKitBuild\obj\FindSafari\Release\BuildLog.htm" FindSafari - 1 error(s), 0 warning(s) -- View this message in context: http://www.nabble.com/Building-Webkit-with-cygwin-failed-tp18189797p1818979… Sent from the Webkit mailing list archive at Nabble.com.

3 6

what triggers the expose event on gtk port
by Luka Napotnik 30 Jun '08

30 Jun '08

Hello. I want to know what exactly triggers the expose event in WebKit. Or, how does GTK+ know when WebKit made changes and needs the expose callback to refresh the screen? Any help appreciated. Greets, Luka

2 3

JSImmediate.h 64 bit compatibility problem?
by Paul Pedriana 30 Jun '08

30 Jun '08

I see the following code, which works for 32 bit but not 64 bit. Is there another file or function somewhere that's for 64 bit? static bool isNegative(const JSValue* v) { ASSERT(isNumber(v)); return reinterpret_cast<uintptr_t>(v) & 0x80000000; }

2 1

What regular expressions do we spend time evaluating?
by Darin Adler 28 Jun '08

28 Jun '08

Since SunSpider time now is about 10% inside the regular expression matcher, it's time to reconsider regular expression optimizations. In particular, we should see if we can find a subset of regular expressions where we can implement a more efficient matching algorithm. I wrote code to dump a histogram of time spent matching regular expressions: <https://bugs.webkit.org/show_bug.cgi?id=19801>. Here's the result from running SunSpider on my computer. The column on the left is the number of seconds spend matching an expression, and the column on the right is the expression: 0.080214 - \b\w+\b 0.079685 - agggtaa[cgt]|[acg]ttaccct (ignore case) 0.077910 - [cgt]gggtaaa|tttaccc[acg] (ignore case) 0.073445 - agggta[cgt]a|t[acg]taccct (ignore case) 0.073439 - a[act]ggtaaa|tttacc[agt]t (ignore case) 0.073387 - agggt[cgt]aa|tt[acg]accct (ignore case) 0.072898 - ag[act]gtaaa|tttac[agt]ct (ignore case) 0.072552 - aggg[acg]aaa|ttt[cgt]ccct (ignore case) 0.072440 - agg[act]taaa|ttta[agt]cct (ignore case) 0.072167 - agggtaaa|tttaccct (ignore case) 0.057528 - >.*\n|\n 0.045544 - "[^"\\\n\r]*"|true|false|null|-?\d+(?:\.\d*)?(:?[eE][+ \-]?\d+)? 0.024901 - (?:^|:|,)(?:\s*\[)+ 0.008831 - ^[a-zA-Z0-9\-\._]+(a)[a-zA-Z0-9\-_]+(\.?[a-zA-Z0-9\-_]*) \.[a-zA-Z]{2,3}$ 0.001060 - ^[\],:{}\s]*$ 0.000517 - \\. 0.000264 - [\0\t\n\v\f\r\xa0'"!-] 0.000158 - !\d\d?\d?! 0.000062 - -.* 0.000026 - ^ 0.000024 - ('|\\) 0.000010 - ^[^-]*- I wish we had a test with some better regular expression coverage! Here's one other case with some significant regexp time in it, I think perhaps from JSON validtion? Loading 280slides.com and looking at the test presentation: 0.028335 - \/\/.*(\r|\n)?|\/\*(?:.|\n|\r)*?\*\/|\w+\b|[+-]?\d+ (([.]\d+)*([eE][+-]?\d+))?|"[^"\\]*(\\.[^"\\]*)*"|'[^'\\]*(\\.[^'\ \]*)*'|\s+|. 0.006968 - \S 0.001098 - "[^"\\\n\r]*"|true|false|null|-?\d+(?:\.\d*)?(?:[eE][+ \-]?\d+)? 0.000422 - \\. 0.000263 - (?:^|:|,)(?:\s*\[)+ 0.000259 - [\:\+\-\*\/\=\<\>\&\|\!\.\%] 0.000245 - [^\s] 0.000193 - [\+\-\*\/\=\<\>\&\|\!\.\[\^\(] 0.000119 - \s 0.000109 - ^[0-9a-zA-Z_\-+~!$\^{}|.%'`#&*]+/[0-9a-zA-Z_\-+~!$ \^{}|.%'`#&*]+\+xml$ (ignore case) 0.000042 - ^\w 0.000023 - ^[\],:{}\s]*$ I had a hard time finding pages and tests where regular expressions took enough time to be worth looking at. -- Darin

2 1

转发: could someone give a hand in porting WebKit to Linux?
by leslie.lg＠hotmail.com 28 Jun '08

28 Jun '08

Hi All, I’m working on porting QT4.4.0 to uclinux these days, and successfully ported GUI, Core, Networks libraries running smoothly on my blackfin 561 ezkit, but I stuck at webkit lib. As Jean-Charles Verdié said, you have webkit run flawlessly on uclinux? How to do that? What modification do I suppose to do with webkit involved in Qt4.4.0? My uclibc is 0.9.29, but with old version of pthread lib, so it doesn’t have pthread_getattr_np or pthread_attr_getstack functions. Indeed, I have tried some thing on it, I paste both of functions mentioned above to collector.cpp, in webkit source tree. And the compilation is passed, but when I run webkit examples on my boad, it always crashes in libQtWebkit.so. Please help me! Thanks & Best regards, Leslie

1 0

Re: [webkit-dev] reclaim of number cells
by Zoltan Herczeg 27 Jun '08

27 Jun '08

Hi Maciej Thank you for your reply. The algorithm itself is very simple. Think about the following expression: a = b + c + d. The result of (b+c) is just a temporay value, which is not used anymore after this expression is evaluated. From technical viewpoint, a shadow register file tracks those numbers cells, which reference counter is exatly one. If a reference counter increases (caused by instructions like mov, put_by_id, call...), we remove the particular number cell from this shadow register file. However, if a new value is stored in a register, and a number cell is pointed by its shadow register, we can safely reclaim that number cell. The perfromance drawback is that we always perfrom this check after some selected instructions (like add, sub, mul, div, mod, ...) even if the current benchmark never allocates a number cell. Your idea of 64 bit VM registers are interesting. That makes this patch a complete waste. Since most of our systems are 64 bit systems, we already have 64 bit VM registers :) That makes me wondering, how do you plan to store a 64 bit pointer using a 64 bit VM register? Cheers, zoltan > Hi Zoltan, > > The performance results are certainly interesting. Would you mind > explaining the algorithm a bit more? > > Also, do you know why it is slowing down the tests that it does? In > general the tests that seem hurt in performance are ones that would > use a lot of immediate numbers. I can understand why they would not > benefit from reducing allocation of number cells, but why do they get > slower? If we understand that, we might be able to come up with a > version that has the upsides but not the downsides of your patch. > > I should also note, we are considering using 64-bit VM registers (with > a special encoding that well let us store a double, a pointer or an > int without a separate type tag, basically all non-double types would > look like a NaN but not the standard NaN, so the low 48 or so bits are > free.) In such a scheme, number cells would be GC-allocated only if > the number is stored in an object (or possibly not at all if we decide > we can make objects and arrays bigger). > > That's probably a few weeks off though, so I'd be interested in > whether we can make your patch work. > > Regards, > Maciej > > On Jun 26, 2008, at 2:38 AM, Zoltan Herczeg wrote: > >> Hi All, >> >> I am Zoltan Herczeg, working for the Dept. of Soft. Eng., Univ. of >> Szeged. >> I am currently investigating the JavaScriptCore engine to find some >> speed-up solutions. During my investigation, I have encountered an >> interesting issue for the Garbage Collector. Namely, in the case of >> some >> sunspider benchmark programs, the GC is frequently called. For >> example, >> the runtime of 3d-cube is 121 msec, and it calls the slow Collect() >> function about 170 times. Thus, the garbage collecting takes the large >> portion of its runtime (~30%). Such benchmarks cunsume the available >> Number Cells in a short time, so the GC has a lot of work to reclaim >> these >> cells. My idea is to have a simple algorithm, which can reclaim many >> of >> these cells without the help of the GC. I have attached an >> experimental >> patch, and the performance-results as well. As you can see, it is a >> rather >> selective patch. Some programs benefit from it (25% performance >> improve >> for 3d-cube), while some do not (like controlflow-recursive, which >> does >> not allocate any number cells at all). Perhaps this approach could >> be a >> custom-speedup, which should be turned on, when a treshold value is >> reached. (Number of collects() in a given time is greater than a >> pre-defined value.) However, the mainloop (Machine:: PrivateExecute) >> should be duplicated in this case, which is not an easy task because >> of >> the computed gotos. Do you have any other idea to make this patch >> useful? >> >> Regards, >> Zoltan<gc- >> improve >> -34737 >> .patch><compare.txt>_______________________________________________ >> webkit-dev mailing list >> webkit-dev(a)lists.webkit.org >> http://lists.webkit.org/mailman/listinfo.cgi/webkit-dev

2 2