Mobile/StartupOptimization: Difference between revisions

 
(4 intermediate revisions by 2 users not shown)
Line 10: Line 10:


Here are some potential optimizations:
Here are some potential optimizations:
* Startup bugs
**
[https://bugzilla.mozilla.org/show_bug.cgi?id=121341]
[https://bugzilla.mozilla.org/show_bug.cgi?id=433098]


* Improve fastload performance
* Improve fastload performance
Line 30: Line 35:


* removing comments from all evaluated source files (js, css, xul).  Less to read in.
* removing comments from all evaluated source files (js, css, xul).  Less to read in.
** Bug needed. Possible preexisting bug.
** Bug 463066 [https://bugzilla.mozilla.org/show_bug.cgi?id=463066]


* look at what system libraries are being opened on start.  strace shows we looking around for system version of nspr, nss, ect?  Maybe we can help the dso loader.
* look at what system libraries are being opened on start.  strace shows we looking around for system version of nspr, nss, ect?  Maybe we can help the dso loader.
Line 39: Line 44:


* merge all XPT files.
* merge all XPT files.
** bug needed
** Bug 469873 [https://bugzilla.mozilla.org/show_bug.cgi?id=469873]


* Can JS files be precomputed/precompiled?
* Can JS files be precomputed/precompiled?
Line 48: Line 53:


* Reduce js files. there are probaby some js completes that we do not need. (default command line handling, for example).
* Reduce js files. there are probaby some js completes that we do not need. (default command line handling, for example).
** bug needed
** Bug 468607 [https://bugzilla.mozilla.org/show_bug.cgi?id=468607]


* Is sqllite doing a bunch of unneeded F_SETLK64s  (fcntl64)?  Maybe we do not have to be so overly protective of other writers?
* Is sqllite doing a bunch of unneeded F_SETLK64s  (fcntl64)?  Maybe we do not have to be so overly protective of other writers?