logs archiveIRC Archive / Freenode / #firefox / 2015 / August / 30 / 1
Philias
By the way, its still going with me watching videos. Initiating some garbage collection runs manually and then reloading the relevant pages seems to have mostly alleviated the symptoms
What the heck, Firefox just tried updating although another instance of it is still running and & it worked :D
Nice one, thanks for that :)
acmehandle
Is it true that mozilla is dumping firefox in favor of using chromes framework thing?
chrome sucks
I wish you guys would consult with me about these kinds of things before you make such rash decisions.
b0at
no
https://wiki.mozilla.org/WebExtensions
it's going to move to an extension api to make multiple processes and moving to the servo rendering engine in the future possible. it's supposed to be a superset of the chrome/opera apis in order to speed things up, but they're working with extension developers to make it not suck.
acmehandle
Ok. This seems ok. I'll have to think on it and let you know if I approve 100%.
I still have some reservations.
b0at
we have plenty of time for forks to appear and decisions to change
and having a permissions model for extensions would be nifty, as well as not having to dig deep in the browser's interface's bowels just to make hotkeys and things work.
Philias
The last time I heard about it, it was a claim that certain add-ons wont work anymore as a result of structural changes, among them DownThemAll (which I like)
b0at
https://wiki.mozilla.org/WebExtensions notes that a downloads api is on the list to be supported, which may allow the most xul-ish parts of dta to work
Philias
Well, I really only ever use it as a download manager (it was the first one I tried, I think, and I grew to like it), but meh, Im still not really fond of Electrolysis
Having multiple processes means having more overhead and more expensive IPC. Linux users should also note that processes arent as cheap on Windows as they are on Linux, where forking is considered totally OK
b0at
it also means better encapsulation and parallel processing (which servo takes advantage of to speed up rendering)
         

Philias
Hm? One can have parallel processing just fine or even with multiple threads (it can be even faster because the overhead is typically lower, as far as I know; it certainly wont be bigger)
I know that improving encapsulation is the motivation behind this, as I should have stated (and once had, before I edited it out ;) ), but my point is rather that I dont really need that and would not like to pay the price for it
b0at
the main thing is to be able to modularize extensions so they don't depend on the interface thread
Philias
Hm, this could be helpful if one of them bugs out
b0at
so the main thing we lose is being able to poke and prod the interface's xul documents in unique ways
but what the webextension's api list shows is that this ability isn't required to do most things extensions do, and mozilla devs are still willing to work to extend it to do the rest
to me, i think it's a good thing this came as late as it did. browsing as a way of working with software has had a chance to mature a little, so i'm not so afraid they're going to waste their time.
Philias
Well, maybe it wont be as bad as I imagine it at the moment :)
And its certainly advantagous to develop things in an environment where its already clear what people want
(and need)
b0at
it could end up being faster and more usable than opera in its heyday. it had a lot of great built-in features like a customizable interface, hotkeys, gestures, and stuff that inspired firefox and its extensions.
not _as_ customizable, but they spent a lot of effort making it usable
too much, perhaps
Philias
Ah, yeah, I can remember people liking it a lot
Its also part of what I liked about Firefox, being able to shift things around in the interface and to use different skins :)
b0at
and there's also the browser.html project and similar that try to re-implement the interface with the new html5 goodies instead of non-standard xul, so we may be able to just stick that on top and get some of our current capabilities back. there's no telling.
though that seems to be geared more to mobile platforms
Philias
But I never went really crazy with it, I fear that maybe I have missed out on something  is there a list of add-ons to try out before they wont work anymore? ;)
Hm
b0at
i wouldn't worry about extensions being a sinking ship until they've passed parity with chrome and fully ported tab mix plus and noscript
Philias
Im not really sure I like that approach (what a surprise :D ), HTML is not exactly the fastest design language around (I mean, browsers got pretty fast, with JS benchmarks showing it to be 1.6 times or so as slow as C even without asm.js, if I remember correctly), but look at how much effort had to be put into optimizing it, for years, and I fear that using the same parser for both, site content and the browser interface itself and then for su
ch a contrived language, could lead to more vulnerabilities
... in the future (dammit)
b0at
as long as they're properly separated, it'll be fine. and yeah, that's another reason to prefer html and standards: the html engine has had a lot more security and performance work than the xul side has
duplicated effort
Philias
Oh, it does? Hm
Err, well, I mean, I supposed one shouldnt really need to sandbox the interface itself, that should be trusted code
But on the other hand, if one is to run potentially untrusted add-ons, maybe a bit of sandboxing wont hurt ;)
It just feels wrong to have HTML as a language for interface definitions, I always looked down at programs using something like XML for it
b0at
xul is xml...
Philias
Yeah, thats what occured to me, too, when I thought about it a bit more ;)
acmehandle
That browser-spawning-multiple-processes is a problem, at least for me. When I would run chrome and firefox on my machine. I found that firefox would get starved for resources if I left my machine running over night at some point. Chrome would hog up the machine resources.
I'm on a linux box btw.
b0at
unless that's developer/aurora or a nightly, it shouldn't even have that turned on yet
acmehandle
so I simply stopped using chrome.
Philias
But Chrome did have it for a long, if not from the beginning
acmehandle
But most most people would assume that it was firefox that was borked just because on the surface chrome was still working but firefox was freezing up
I realized what was going on.
and it really made me angry at chrome
         

Philias
It could also lead to people being more lenient about resource leaks, if they know (or at least think they do) that they will be cleaned up soon anyway
acmehandle
I dont know, any well developed app should not display memory leaks of that magnitude
thats microsoft level resource leakage
if anything I think google developed it that way on purpose.
and even if not its still shameful of google to release garbage like that.
Philias
Hm, I mean, there has been put much work into the memory subsystem of Firefox in the past (you probably remember the times of Firefox 4 and the subsequent releases, when it was really bad), but it got nice again (mostly), so if this work can be kept, I dont think it would get that bad
acmehandle
I agree, I just dont remember myself being as angry at mozilla in the past as I am at google for their chrome. But rephrase what I said above. I dont think this boils down to some resource leakage. Its just that every other application I was running was fine. Its that firefox was the one that became slow as hell and very often simply freezing up. While chrome was just moving along. There was something very parasitic about
the whole thing. So I avoid using chrome and use firefox and its works much better when I have a bajillion tabs open than chrome in a comparable situation. Chrome spawns process after process after process for each tab. I saw them when I ran ps aux.
Philias
Well, I never used Chrome, so I guess cant add anything of value here. Id just like you to know that programmers in general (as to my impression) usually dont try for their programs to be as parasitic as possible and grab resources for no reason, but instead are either lazy enough not to care (or even if they do, it can be hard at times to keep resource consumption low) or are just trying to squeeze out the most in terms of performance.
If this means trading memory for performance, its usually done plainly for the reason that spare memory in itself doesnt really benefit anyone. Of course this train of thought doesnt hold if they were really up to messing with Fx as much as possible in very intricate ways, but & lets just hope they arent, we could be in big trouble if Google was not just somewhat evil anymore but got really insidious and nasty instead ;)
acmehandle
That is a good point as well.
I am just astounded at how bad chrome was and how it affected firefox so significantly.
Considering that I execpted *much* better from google I cant help but conclude that there was something nefarious going on
and if their programmers are *that* lazy and indifferent then what the heck is google getting all those billions for?
Philias
Nah, I wouldnt jump to conclusions so easily, it may just have been a result of both programs in total taking up a lot of memory, more than your computer had to offer, and Chrome winning up on it (maybe because Windows decided to give it more, because it was made up of multiple processes  there surely isnt a special case in it which would avoid that)
They get them for their success, not for how good their programmers are, to say it indirectly
acmehandle
I'm on linux, not windows
Philias
Oh, sorry
Then maybe you should have set up cgroups properly
:P
No excuses for you in that case ;)
acmehandle
either way, chrome is sloppy at best.
Philias
Anyways, its more of a trope that the bad stuff always seems to win over the good  but usually more expensive or harder to use  stuff
It happened many times in history and perhaps most prominently in tech
acmehandle
Good or bad, it is what it is. I stopped using chrome and am pretty happy.
Philias
So am I and I will be even in the future if people dont break Fx with this move ;)
*-even
Oh, and if they could maybe fix that weird issue I talked about, that would be nice
If nobody objects, Ill be leaving in maybe ten to twenty minutes, you should have spoken up by then or youll probably have to wait until later this day or even tomorrow ;)
purpleidea
fwiw this link reliably crashes firefox v39.0 on ubuntu: https://www.quantamagazine.org/20150828-john-conway-a-life-in-games/
if anyone with a firefox bz account can report this (assuming they can confirm the crash) then i'd appreciate it, thanks ^^^
auscompgeek
purpleidea: latest version is firefox 40.0.3
Peng
39 is ESR, I think?
auscompgeek
I just double-checked, ESR is 38
Peng
oh
:<
purpleidea
auscompgeek: unfortunately on this machine i'm not on latest ff, but figured i'd report the link here in case someone can easily reproduce this.
auscompgeek
purpleidea: I can't reproduce it in dev edition, fwiw.
purpleidea
auscompgeek: okay, thanks for letting me know`
nor can i in 39.0.3 on a fedora machine, fwiw
anddam
hello, I saved some bookmarks with tags in order to use them as shortcuts
now if I select such a bookmark in Show All Bookmarks window then the "Tags" in bookmark's details downward the page doesn't show any tag at all
yet the shortcut keeps working
« prev 1 2 next »