Re: Linux 2001-05-10 gtkEmbed smoketest results

[Curt Patrick]

You need a special build of gtkEmbed.  The build I make daily is found under /u/curt/reflow/cm/mozilla.  The .mozconfig file should look like:

ac_add_options --enable-debug
ac_add_options --disable-tests
ac_add_options --enable-cpp-rtti
ac_add_options --enable-reflow_perf

Also, you need to modify mozilla/layout/html//base/src/nsPresShell.cpp such that the line which currently contains the string "SetDumpFrameCounts(dumpFrameCounts)" is changed to "SetDumpFrameCounts(PR_TRUE)".  It seems we can't control this with prefs in gtkEmbed.  It may be that you can get more detail by setting the SetDumpFrameByFrameCounts to true, but Rod Spears would know about more about that.  If you want more details, or help in running the gtkEmbed against our test set of URLs, give me a call:  x2401.  Otherwise, my scripts are available under /u/curt/reflow/scripts/test/launch_reflow_test.csh.

BTW, closer inspection of the url by url data indicates that this jump in reflows may have taken place yesterday but was masked by an invalid zero reading on the worst url in the list.  Thus, I would say that any checkins between 1:00 a.m. Tuesday the 8th and 1:00 a.m. this morning are candidates for this jump in reflows.

Thanks,

Curt
 

Yung-Fong Tang wrote:

can you point me to the document about the reflow test? how can I run it ?

Cathleen Wang wrote:

Hook, reflow counts went up this morning.  Please review your check-ins.  We need help nail this one.

Thanks!
Cathleen

Tracy Walker wrote:

Linux 2001-05-10 gtkEmbed smoketest results

- Embedding: PASS:

    - Bug #  70714 : Visited link not changing color

    - Bug #  66020 : add CJK/intl support

Memory test results here . test still running (I'll post notice if anything appears unusual)

Reflow test results here . reflows Jumped dramitcally

Leak test results here.