- Dillo-dev - lists.mailman3.com

dillo CSS and charset

by furaisanjin＠gmail.com

Hi. And charset conversion works fine. Regards, furaisanjin

Jan. 22, 2009

dillo CSS and charset

by Johannes.Hofmann＠gmx.de

On Thu, Jan 22, 2009 at 08:34:32AM -0300, Jorge Arellano Cid wrote: > On Thu, Jan 22, 2009 at 10:22:39AM +0100, Hofmann Johannes wrote: > > Hi, > > > > On Wed, Jan 21, 2009 at 07:06:06PM -0300, Jorge Arellano Cid wrote: > > > On Wed, Jan 21, 2009 at 08:29:34PM +0900, furaisanjin wrote: > > > > Hi. > > > > > > > > This is work around. > > > > > > This so called "work around" works very well. > > > > > > While working on this, I used the testcases in the following table: > > > > > > > > > load Reload Back > > > ---------------------------- > > > H__ OK OK OK H = has charset in HTTP header > > > HM_ OK OK OK M = has charset in META > > > H_C OK OK OK C = has remote CSS stylesheet > > > HMC OK OK OK _ = disabled > > > ___ OK OK OK > > > _M_ OK OK OK > > > __C OK OK OK > > > _MC OK OK OK > > > ---------------------------- > > > > > > Both patches pass all of them! > > > > > > I committed a similar approach that has a single code path for > > > calling a_Decode_charset_init(). > > > > > > Please test and comment. > > > > Starting with revision d29cdb5b842e I get repeatable crashes when > > clicking on the "Home" link in about:splash. > > Please try the current tip. Yes, it's fixed! Cheers, Johannes

Jan. 22, 2009

dillo CSS and charset

by jcid＠dillo.org

On Thu, Jan 22, 2009 at 10:22:39AM +0100, Hofmann Johannes wrote: > Hi, > > On Wed, Jan 21, 2009 at 07:06:06PM -0300, Jorge Arellano Cid wrote: > > On Wed, Jan 21, 2009 at 08:29:34PM +0900, furaisanjin wrote: > > > Hi. > > > > > > This is work around. > > > > This so called "work around" works very well. > > > > While working on this, I used the testcases in the following table: > > > > > > load Reload Back > > ---------------------------- > > H__ OK OK OK H = has charset in HTTP header > > HM_ OK OK OK M = has charset in META > > H_C OK OK OK C = has remote CSS stylesheet > > HMC OK OK OK _ = disabled > > ___ OK OK OK > > _M_ OK OK OK > > __C OK OK OK > > _MC OK OK OK > > ---------------------------- > > > > Both patches pass all of them! > > > > I committed a similar approach that has a single code path for > > calling a_Decode_charset_init(). > > > > Please test and comment. > > Starting with revision d29cdb5b842e I get repeatable crashes when > clicking on the "Home" link in about:splash. Please try the current tip. -- Cheers Jorge.-

Jan. 22, 2009

dillo CSS and charset

by Johannes.Hofmann＠gmx.de

Hi, On Wed, Jan 21, 2009 at 07:06:06PM -0300, Jorge Arellano Cid wrote: > On Wed, Jan 21, 2009 at 08:29:34PM +0900, furaisanjin wrote: > > Hi. > > > > This is work around. > > This so called "work around" works very well. > > While working on this, I used the testcases in the following table: > > > load Reload Back > ---------------------------- > H__ OK OK OK H = has charset in HTTP header > HM_ OK OK OK M = has charset in META > H_C OK OK OK C = has remote CSS stylesheet > HMC OK OK OK _ = disabled > ___ OK OK OK > _M_ OK OK OK > __C OK OK OK > _MC OK OK OK > ---------------------------- > > Both patches pass all of them! > > I committed a similar approach that has a single code path for > calling a_Decode_charset_init(). > > Please test and comment. Starting with revision d29cdb5b842e I get repeatable crashes when clicking on the "Home" link in about:splash. Cheers, Johannes Program received signal SIGBUS, Bus error. DilloHtml::getCurTagLineNumber (this=0x28778500) at html.cc:596 596 if (p[i] == '\n') (gdb) bt #0 DilloHtml::getCurTagLineNumber (this=0x28778500) at html.cc:596 #1 0x08063e38 in DilloHtml::bugMessage (this=0x28778500, format=0x80f8198 "HEAD section lacks the TITLE element\n") at html.cc:149 #2 0x08063f2a in Html_tag_close_head (html=0x28778500, TagIdx=32) at html.cc:1569 #3 0x080684de in Html_tag_cleanup_at_close (html=0x28778500, TagIdx=32) at html.cc:1306 #4 0x08068957 in DilloHtml::finishParsing (this=0x28778500, ClientKey=2) at html.cc:627 #5 0x0805a566 in Cache_process_queue (entry=0x287418c0) at cache.c:1153 #6 0x0805b378 in a_Cache_process_dbuf (Op=2, buf=0x0, buf_size=0, Url=0x28741c80) at cache.c:886 #7 0x0805c871 in a_Capi_ccc (Op=4, Branch=2, Dir=1, Info=0x2874fd60, Data1=0x0, Data2=0x0) at capi.c:618 #8 0x08058288 in a_Chain_fcb (Op=0, Info=0x287a3000, Data1=0x0, Data2=0x0) at chain.c:112 #9 0x08078cee in a_Dpi_ccc (Op=4, Branch=2, Dir=1, Info=0x2874f140, Data1=0x0, Data2=0x0) at dpi.c:671 #10 0x08058288 in a_Chain_fcb (Op=0, Info=0x287a3000, Data1=0x0, Data2=0x0) at chain.c:112 #11 0x0807969e in a_IO_ccc (Op=4, Branch=2, Dir=1, Info=0x2874f6c0, Data1=0x2874f8c0, Data2=0x0) ---Type <return> to continue, or q <return> to quit--- ---Type <return> to continue, or q <return> to quit--- at IO.c:418 #12 0x08079980 in IO_callback (io=0x2874f8c0) at IO.c:211 #13 0x080799ff in IO_fd_read_cb (fd=7, data=0x2) at IO.c:287 #14 0x080ca465 in fltk::wait () #15 0x080ca615 in fltk::run () #16 0x0804f03a in main (argc=1, argv=0xbfbff864) at dillo.cc:334 (gdb)

Jan. 22, 2009

[bug]: hang in Css_parse_delaration()

by onepoint＠starurchin.org

On Wed, Jan 21, 2009 at 09:41:45PM +0100, Hofmann Johannes wrote: > On Wed, Jan 21, 2009 at 04:49:27PM +0000, Jeremy Henty wrote: > > > > A simple example that hangs the CSS parser. > Please check current tip. It's fixed! > The page is still not looking pretty, That's OK. Dillo is not responsible for the ugliness of the world! ;-) Regards, Jeremy Henty

Jan. 22, 2009

dillo CSS and charset

by jcid＠dillo.org

On Wed, Jan 21, 2009 at 08:29:34PM +0900, furaisanjin wrote: > Hi. > > This is work around. This so called "work around" works very well. While working on this, I used the testcases in the following table: load Reload Back ---------------------------- H__ OK OK OK H = has charset in HTTP header HM_ OK OK OK M = has charset in META H_C OK OK OK C = has remote CSS stylesheet HMC OK OK OK _ = disabled ___ OK OK OK _M_ OK OK OK __C OK OK OK _MC OK OK OK ---------------------------- Both patches pass all of them! I committed a similar approach that has a single code path for calling a_Decode_charset_init(). Please test and comment. -- Cheers Jorge.-

Jan. 21, 2009

[bug]: hang in Css_parse_delaration()

by Johannes.Hofmann＠gmx.de

On Wed, Jan 21, 2009 at 04:49:27PM +0000, Jeremy Henty wrote: > > A simple example that hangs the CSS parser. The contents of the css > block are the critical thing: it doesn't seem to matter what the > selector is but changing either the parameter or the value inside the > block stops the hang. After aborting the stack trace varies, but > always seems to be inside Css_parse_declaration(). (I've attached > some example stacks.) > > The real-world page that triggered the bug: > > http://dialinf.wordpress.com/2009/01/21/ultraproducts-the-category-theoreti… Please check current tip. The page is still not looking pretty, but it no longer hangs. Cheers, Johannes

Jan. 21, 2009

[bug]: hang in Css_parse_delaration()

by onepoint＠starurchin.org

A simple example that hangs the CSS parser. The contents of the css block are the critical thing: it doesn't seem to matter what the selector is but changing either the parameter or the value inside the block stops the hang. After aborting the stack trace varies, but always seems to be inside Css_parse_declaration(). (I've attached some example stacks.) The real-world page that triggered the bug: http://dialinf.wordpress.com/2009/01/21/ultraproducts-the-category-theoreti… Regards, Jeremy Henty

Jan. 21, 2009

dillo CSS and charset

by furaisanjin＠gmail.com

Hi. This is work around. 1) HTTP server and meta have charset. One example is http://www.so-net.ne.jp/news/cgi-bin/article.cgi?gid=mai&aid=20090119-570-O… diff -r a0230b2a4a32 src/cache.c --- a/src/cache.c Fri Jan 16 14:45:17 2009 -0300 +++ b/src/cache.c Wed Jan 21 20:21:49 2009 +0900 @@ -628,6 +628,7 @@ { char *header = entry->Header->str; char *Length, *Type, *location_str, *encoding; + char *charset; #ifndef DISABLE_COOKIES Dlist *Cookies; #endif @@ -753,6 +754,13 @@ entry->TypeHdr = Type; _MSG("TypeHdr {%s} {%s}\n", Type, URL_STR(entry->Url)); _MSG("TypeMeta {%s}\n", entry->TypeMeta); + a_Misc_parse_content_type(Type, NULL, NULL, &charset); + if (charset) { + entry->CharsetDecoder = a_Decode_charset_init(charset); + if (entry->CharsetDecoder) + entry->UTF8Data = dStr_new(""); + dFree(charset); + } } } 2) Only meta has charset. One example is http://www.picfun.com/app00.html diff -r a0230b2a4a32 src/html.cc --- a/src/html.cc Fri Jan 16 14:45:17 2009 -0300 +++ b/src/html.cc Wed Jan 21 20:22:00 2009 +0900 @@ -2789,6 +2789,7 @@ * this code in another bw might have already changed it for us. */ if (a_Misc_content_type_cmp(html->content_type, new_content)) { + html->stop_parser = true; html->repush_after_head = true; } } Regards, furaisanjin

Jan. 21, 2009

Crashes in dw::core::Layout::createImgbuf()

by Johannes.Hofmann＠gmx.de

Hi Jeremy, On Tue, Jan 20, 2009 at 06:20:32PM +0000, Jeremy Henty wrote: > > For the last few days I have been getting occasional crashes inside > dw::core::Layout::createImgbuf() when opening a new tab. It does not > seem to be predictable: if I restart Dillo and reopen all the URLs I > was just looking at it is fine. > > The gdb backtrace often shows ridiculous values for this: eg. the > attached example has this=0x6. I've seen other impossibly small > values for this, but sometimes the value seems more sensible. > > Any ideas? I'd like to investigate further but I don't know what to > look for. Pointers are welcome! I've never seen these errors yet. Can you try to use "hg bisect" to find the changeset where the error was introduced? I know it's difficult with sporadic errors... Or maybe try valgrind? Regards, Johannes

Jan. 20, 2009