On Aug 18, 12:34 pm, Bob Hassinger <
r.hassin...@ieee.org> wrote:
> "I have a browser crashing
> issue. When I open certain sites like Gmail and Meebo, Firefox slows
> to a crawl and crashes"
>
> Watching all this discussion go by for many months, it seems a key
> root is "the Gmail problem". Do I recall correctly that back around
> 1.2-1.3 that there were a lot of complaints about conflicts between FB
> and sites like Gmail, and a lot of the attention to activation schemes
> flowed from there?
I wish it were so simple.
In FF 2.0, gmail + firebug meant crashes. Eventually we learned that
the problem was much worse: google's pageads code crashed firefox
because of a combo of JS debugging and GC. Fixed in FF3.0
Issues for FF3.0 are all around resources. (I discount the "crashes"
in the report above, they may not mean the same thing I do, it may not
be Firebug, etc). Gmail is a common reason for folks to want a
browser that is both a debugger and user tool. But it's not the only
reason.
>
> Likewise, as the above quote reflects a lot of the current concern
> would go away if there was not a need to be off/inactive/whatever for
> Gmail and maybe some fairly limited set of other sites.
I don't agree for two reasons.
First controlling the resource requirements of Firebug will always be
needed. It is not specific to gmail or even to trying to combine
debugging and other web activity. A serious debug tool will be applied
to serious applications and they will stress the limits of resources.
We need to be able to control Firebug.
Second, its not really gmail and a limited set of sites. Rather gmail
and similar sites are just the vanguard of applications that will
continue to roll out. We can't cut and run, we have to learn how to
defeat these problems.
>
> My question is how well "the Gmail problem" is understood and what
> prospects and barriers might exist to address it?
Yes, generally the problem with gmail, and similar AJAX applications,
is that firefox is not designed for debugging dynamic code and these
applications compile a lot of code dynamically. I wrote a lot of
Javascript that does some wacky stuff to make up the gap. It works
but it is slow.
>
> That subject seems pretty obvious so I would assume it is a hard
> problem for some reason. Otherwise it would have been fixed long
> ago. But whatever is going on to cause it would seem to be a
> fundimentsl flaw. To be effective and solid FB should be able to work
> on all sites without needing to worry. It would seem Gmail is doing
> something legitimate that FB can not deal with effectively in the
> current design.
Yes your analysis is quite correct.
>
> Do the new tools in other browsers that have been mentioned have
> similar problems? If not, why?
I don't know. Prior to my work with Firebug I don't believe that there
was any way to debug these applications. However, new Firebug-like
debuggers are now out in IE8 and Opera; V8 is coming along. I like to
know how well they work.
>
> This issue seems significant enough to even justify getting support or
> changes in FF itself if that is where it has to come from.
I've been working on this and making steady progress. There are two
aspects.
First, what technical solution makes sense? Here we don't have much to
work with. I'm the expert on the firebug side, there is no one on the
Firefox side. There are experts deeper into the Javascript engine.
Second, who can do the work? I have a learned enough to create C++
code for making some improvements, but this work is quite time
consuming. Fortunately Firefox team has expressed a lot of support for
Firebug work. We've agreed to first focus on adding Firebug to nightly
tests of Firefox then put it on the Firefox "you must pass this test
before you move one" list. At that point we can work on Firebug on
Firefox trunk and on Firefox trunk with Firebug installed.
(To be sure there are a lot of other opportunities to exploit if we
get this done).
>
> Just think how much more time the FB developers would have to do
> useful work if the problem and the endless discussion about it could
> go away ;-)
I share your concern that resources we put into activation prevent us
from working on other problems. However in a lot of ways the 1.4 work
was a great investment. Without setting out to do so, we plowed
through many obscure corners of Firebug and rewrote some complex code
we did not understand before. Perhaps it will take 1.5 to iron out all
of the wrinkles, but at least we know how many pockets the shirt
has ...
jjb