Closed Bug 468486 Opened 16 years ago Closed 16 years ago

Leak & Bloat tests should not start up with the What's New page

Categories

(MailNews Core :: Build Config, defect)

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED
Thunderbird 3.0b2

People

(Reporter: standard8, Assigned: standard8)

Details

Attachments

(1 file)

Attached patch The fixSplinter Review
The leak and bloat tests currently have a pref:

user_pref("mailnews.start_page_override.mstone", "1.9pre");

The value of this is wrong, and really we should just set it to "ignore" at this stage because:

a) The what's new page only gets shown on the Refcnt tests, not the malloc ones (because we currently don't reset the profile in-between).
b) The content of the what's new page may change over time and cause us issues.

This shouldn't affect any of the current figures when it lands (due to not being included in the trace-malloc tests).
Attachment #351932 - Flags: review?(bienvenu)
Attachment #351932 - Flags: review?(bienvenu) → review+
Checked in: http://hg.mozilla.org/comm-central/rev/ff348d4f7817
Status: NEW → RESOLVED
Closed: 16 years ago
Flags: in-testsuite+
Resolution: --- → FIXED
(In reply to comment #0)
> b) The content of the what's new page may change over time and cause us issues.

I'd argue that if the first-run page causes leak(s) thats a problem.  Though one could argue that if it leaks, it would be a Core Gecko leak that Platform would need to worry about anyway... if nothing else we should *manually* check the firstrun page startup against leaks for each platform, on each update that uses one.

(Probably a different bug fodder, this was just easier :-) )
(In reply to comment #2)
> (In reply to comment #0)
> > b) The content of the what's new page may change over time and cause us issues.
> 
> I'd argue that if the first-run page causes leak(s) thats a problem.  Though
> one could argue that if it leaks, it would be a Core Gecko leak that Platform
> would need to worry about anyway... if nothing else we should *manually* check
> the firstrun page startup against leaks for each platform, on each update that
> uses one.
> 
> (Probably a different bug fodder, this was just easier :-) )

Given the work being done in Firefox to prevent leaks caused by pages, I doubt whether we have any leaks from the actual page. This will most likely affect total mallocs/heap.

I will probably consider at a later stage adding the firstrun page back in, but if I do, then it will be served by a local http server (possibly the fake one) and static in content so that we don't get random changes.

I'm certainly thinking we potentially have bigger problems in MailNews to resolve, and I'd rather focus getting our tests running with those than duplicating effort on a very small part, where Firefox is already covering a large part of it.
You need to log in before you can comment on or make changes to this bug.