User:Rcampbell/1.9/3.0 Planning Discussion notes

Notes gathered with a specific interest on testing

Web compatibility issues, top 50 incompatibilities should be considered and areas where standards break.

Gather performance differences between previous browser versions.

Image cache performance comparisons. Gather stats.

dbaron suggests varying browser tests adding more randomness.

stuart suggests turning off page cache entirely.

davel suggests multiple performance test suites.

Pav: Cairo remains P1, Stuart: Cairo P1 or P2

davel: martijn has test cases that could cover cairo

pav: cocoa widgets on mac... easier to do cairo with cocoa widgets. cocoa widgets will be cleaner than quickdraw

roc joins discussion on phone

roc: jprof running on tinderboxes to check performance data for easier validation

boris: fix scrolling issues?

vlad: linking content and chrome linking as P1? probably P3...

roc: need to look at cost:benefit ratio. bump linking to P2.

boris: fix test frame for regression

pav: need to fix in linux, xft vs pango...

vlad: regressions become P1 from frame display list

boris: fixes should be minor

vlad: units fixing in gecko

roc: eli has patch, depends on cairo because of scaling nastiness

vlad: issue with gfx pixel alignment

vlad: reflow issue for compatibility, P2?

pav, roc: may be blocking other fixes

vlad: ACID2... P2 for web-compatibility, P1 within his block

schrep: we should be able to re prioritize work in progress. If something is going to take much more time and is at a lower priority, we should be able to move onto something else.

dbaron: resets are costly too.

vlad: caret painting rewrite, regression P1 - already part of cairo rewrite.

vlad: residual style handling, P3 - breaks some sites

...

vlad: DOM piece in danger of being cut

brendan says it got hairy

vlad: language-agnostic DOM has landed, feature's in.

meeting begins to skew from priorities about here

boris: some regressions left. crash on EMRestart. during shutdown trying to compile a XUL script. crash on corrupt fastload file?

jonas: should be able to handle malloc failure.

dbaron: reduce tolerance to leak regressions

vlad: talk to mark and see what his plan is

boris: fastload thing he will need help

vlad: SVG image support, P3

... leave to grab laptop power

TP: time-to-load tests

TS: startup time

--Rcampbell 11:26, 28 June 2006 (PDT)