Page 2 of 2

Re: please help to reproduce Firefox 3+ bug 460880

Posted: Thu Jul 02, 2009 11:18 pm
by al_9x
GµårÐïåñ wrote:Sorry for the delayed response, I will go ahead and drop it in my XP VM and test it at various levels, I am thinking current, current+1 import (x2), and finally x2+import (x3) and get back to you, give me a couple of hours, I just got home.
Did you get a chance to try it?

Re: please help to reproduce Firefox 3+ bug 460880

Posted: Thu Jul 02, 2009 11:27 pm
by GµårÐïåñ
Yes I did and I was actually up till 3 AM (as Tom can verify) but I couldn't reproduce the problem. I tried as many variations of the problem as I could conceive but nada. I tried in 3.6pre and 3.5 but I was thinking about trying it with 3.0.11, been working and hadn't had a chance to get to it, will have to wait until I get home. Anything else you want to tell me so I can try?

Re: please help to reproduce Firefox 3+ bug 460880

Posted: Fri Jul 03, 2009 12:13 am
by al_9x
GµårÐïåñ wrote:Yes I did and I was actually up till 3 AM (as Tom can verify) but I couldn't reproduce the problem. I tried as many variations of the problem as I could conceive but nada. I tried in 3.6pre and 3.5 but I was thinking about trying it with 3.0.11, been working and hadn't had a chance to get to it, will have to wait until I get home. Anything else you want to tell me so I can try?
Thank you both for your efforts, didn't mean for this too keep you up at night.

So just to be sure we are in sync, you grew your 4MB places.sqlite (to what size? how many bookmarks?), dropped it into a 3.5 profile on a 32 bit XPSP3 VM, executed the bookmarklet from the address bar several times, and every time all the tabs opened? Is that correct? Popup blocking was on? What is your CPU? (I think threading/cores may play a role here) And your host os?

The only thing I can think of to ask you is, if you have access to an XP32 SP3 machine, to try it there. Both therube and I can get it to fail and that's what we are running. Perhaps therube has some additional pointers.

therube, can you please post your CPU, mine is 3gHz p4 hyperthreaded.

Re: please help to reproduce Firefox 3+ bug 460880

Posted: Fri Jul 03, 2009 12:29 am
by therube
In my case, my places.sqlite would have had (virtually) no history, but it did have 21K bookmarks.

I was on an AMD Sempron 3000, 512KB memory, XP SP3 x86, & I was running FF35.

It was easy for me to duplicate your issue.
It was hard for me to have it work as expected, & it was only those two times that it did work.

Re: please help to reproduce Firefox 3+ bug 460880

Posted: Fri Jul 03, 2009 12:41 am
by al_9x
therube wrote:I was on an AMD Sempron 3000
This one is single core?

Re: please help to reproduce Firefox 3+ bug 460880

Posted: Fri Jul 03, 2009 12:57 am
by GµårÐïåñ
al_9x wrote:Thank you both for your efforts, didn't mean for this too keep you up at night.
No worries, you didn't keep me up, I am a night owl and since I get to do most of my "testing" and stuff after I get home from work, I end up staying up really late, catch a couple of hours of sleep and then back up at 6:30 AM and head out. The longer my to-do list, the longer I have to stay up to get them done, nothing specifically to do with you and glad to do it.
So just to be sure we are in sync, you grew your 4MB places.sqlite (to what size?), dropped it into a 3.5 profile on a 32 bit XPSP3 VM, executed the bookmarklet from the address bar several times, and every time all the tabs opened? Is that correct? Popup blocking was on? What is your CPU? (I think threading/cores may play a role here) And your host os?
I tried with original places at 4 MB and also with one import which brought it up to about 5.75 MB (not sure why it didn't just double) and also with two imports (approximately 11.3 MB) which was slow ass but functional. It was tested on XP SP3 for all of them, the 3.6pre with latest nightly, 3.5 official release, for a total of 6 test cases. The VM is running on a P4/4GB RAM host with 512 RAM assigned to the VM. Popups were at their default setting (on) and I think that covers everything. I was going to try it on 3.0.11 with the 3 sets tonight and see how that goes.
The only thing I can think of to ask you is, if you have access to an XP32 machine, to try it there. Both therube and I can get it to fail and that's what we are running. Perhaps therube has some additional pointers.

therube, can you please post your CPU, mine is 3gHz p4 hyperthreaded.
I AM running on 32 bit and the VM is also 32 bit and my P4 HT machine is 3.20 Ghz with 4 GB RAM, so not sure what else I can do to change that unless you want me to create a VM that is 64 bit and that's going to take me MUCH longer to test since I have to build it first.

Re: please help to reproduce Firefox 3+ bug 460880

Posted: Fri Jul 03, 2009 1:02 am
by therube
al_9x wrote:
therube wrote:I was on an AMD Sempron 3000+
This one is single core?
Perhaps better thought as a half-core :lol:.
Yes, a relatively ancient single core CPU.

Re: please help to reproduce Firefox 3+ bug 460880

Posted: Fri Jul 03, 2009 1:25 am
by GµårÐïåñ
al_9x wrote:
therube wrote:I was on an AMD Sempron 3000
This one is single core?
Dude you kidding? No offense to therube or you but this is a relatively ancient AMD even by their standards, as single core as it gets. :P

Re: please help to reproduce Firefox 3+ bug 460880

Posted: Fri Jul 03, 2009 1:30 am
by al_9x
GµårÐïåñ wrote:unless you want me to create a VM that is 64 bit and that's going to take me MUCH longer to test since I have to build it first.
no there is no point in 64. therube and I are both on 32 bit xp sp3 direct (no vm). As I mentioned, the only thing you can do to get closer to our environment is to try on XP32 not in the vm. You are running 32bit win7, right? What VM vendor? Obviously I am not asking you to install xp32 sp3, but if you do somehow come across such a machine maybe you can try it.

And finally I just want to confirm, since when your are close to the threshold the bookmarklets sometimes work, you issued the keyword command several times and each time all the tabs opened?

Re: please help to reproduce Firefox 3+ bug 460880

Posted: Fri Jul 03, 2009 6:54 am
by GµårÐïåñ
al_9x wrote:no there is no point in 64. therube and I are both on 32 bit xp sp3 direct (no vm). As I mentioned, the only thing you can do to get closer to our environment is to try on XP32 not in the vm. You are running 32bit win7, right? What VM vendor? Obviously I am not asking you to install xp32 sp3, but if you do somehow come across such a machine maybe you can try it.
Ok. I see what you are saying but unfortunately the only XP I have access to at the moment is a VM, so is my Ubuntu, and yes I use Win7 primarily and have access to a Vista Ultimate and Business editions on two laptops. I will ask two of my developers who each keep an XP box for various hacks and tweaks that require legacy drivers. I will ask them to try it if it has internet access or if not ask them to see if they can setup for testing purposes, will let you know tomorrow. Also being developers as well, they might be able to offer another perspective.
And finally I just want to confirm, since when your are close to the threshold the bookmarklets sometimes work, you issued the keyword command several times and each time all the tabs opened?
Yes, BUT at the 3x level it was much slower as I stated earlier and sometimes it would "time out" if you will and show the neterror page but a simple click on the "try again" button would load the page. I am going to put that up to potential socket or canvas overflow, not sure though.