[arch-general] usable browser?

[Rogutės Sparnuotos]

Arvid Picciani (2009-11-27 08:02):
> Rogutės Sparnuotos wrote:
> >With what you wrote above - no, no options.
> >
> >It sounds like you dump software as soon as you encounter any annoyance.
> 
> point, sadly the annoyances usually come in large bulk as feature
> "improvements" together with crucial updates, i actually want. Hence
> i figured i want a browsers that is NOT based on the idea to make
> everything WORSE.

Well, if every upgrade brings 4 new features (1 for me, 1 for you, 1 for
developers, 1 for marketing dept.), we have a universal and bloaty piece
of software. And there's not much choice but to adapt, because browsers
employ too much tangled-together technologies to stay small.

> >Wouldn't it be wiser to workaround them, since annoyances (or small bugs)
> >are always part of everything?
> 
> i do that up so some degree where the workaround consumes more time
> then    stealing my girlfriends mac. Ie the ff workaround was
> avarage 1 day fixing time each update since they managed to
> introduce workarounds for my workarounds.

Does the Mac require no maintenance? Is it because you buy some 10.x
version and there are no major updates afterwards? Isn't the same with
something like Ubuntu LTS, where you install it with e.g. Firefox 2.x and
stay with it for three years (and you get to know it's bugs during that
time, and your workarounds aren't worked around)?

> >What kind of I/O activity do you see with Firefox?
> 
> disk i/o. It's flush() in a busy loop, says kernel wakup debuging
> mode. well my kernel  debugging skills are limited.
> i solved it by sticking .mozilla on a ram disk. that worked until
> the next memory leak bug, then kswapd died out the disk. I tried
> then mounting .mozilla to vaporspace but it would just make ff crash
> constantly, so i gave up.
> 
> How do you measure it?
> 
> iotop. powertop. strace.

powertop doesn't complain on this server-ish desktop. I'll try iotop some
other time.

> >I don't see any problems on my side.
> 
> no one does. the bug got rejected as "can't reproduce". which
> propably means "buy a bigger disk faggot. everyone nowadays runs
> kde/vista/whatever" bleh...

"Not reproducable" usually means exactly what it says. And if the assignee
can't reproduce it, you usually have to take initiative and start
debugging, asking for help in the bug report on the way. Anyway, Mozilla's
bugzilla is a huge and not too warm a place - you need a lot of patience
there.

> ff always used an entire core, which i care less about because i
> have another, but since i use chrome i got used to leaving my
> browser open.

Here, Firefox (dressed up with Vimperator) is always open in the
background. It is not too shy on memory and very slow to start up, but
I see no disk/cpu activity even when I'm browsing (39 open tabs,
javascript disabled with NoScript - things change when I start playing
flash movies, running sites with bad javascript). Something must be wrong
in your system. Did you try a LiveCD of some other distribution? Did you
try binaries from mozilla.org? Did you try disabling Firefox's on-disk
cache? Do you see the same symptoms with a clean profile and a clean
session (no tabs open)? Do your problems happen only on particular sites?

> 
> oh did i mention firefox now depends on dbus?
> Call me whatever you want to, but i actively refuse to run any
> software that starts user space dameons that starts user space
> dameons that start  a power consuming poll loop on my bluetooth
> device until either laptop or my mobile phone die.

I have something against dbus too, but that something is superstitious
because I used Linux when dbus wasn't there and still don't see any good
that it brings (but neither have I tried to find out).

> IgnorePkg   = dbus dbus-core gconf dbus-glib

A strange line you have here. It's ok as long as you use it to catch the
packages that depend on dbus and recompile them, otherwise you couldn't
use qt, xulrunner, ... But recompiling them just to remove the dbus
dependency is superfluous, because having dbus installed doesn't mean that
it will be used or that the daemon will be needed.

Even though Arch's xulrunner is compiled with dbus support, I think that
Firefox doesn't use it yet and I believe that for now it is only made
available for extensions (perhaps with plans to use it for notifications
and talking to the network manager later).

Anyway, no dbus daemons are running here, even though all the dbus*
packages are installed as dependencies.

> >What sites were incorrectly rendered with webkit?
> 
> ebay.de did. now it works. dunno who fixed it.
> but you got a point there, it's been a while since i tried webkit.
> maybe it improved significantly after chrome opensource'd. I'll try
> one of these webking thingies again. suggestions?  um actually i
> know, uzbl. will report back if it still sucks as much as it did a
> few months ago.

I'm using surf besides Firefox. Don't know what usage patterns it was made
for, but I like it when I want to sign on e.g. google.com or don't trust
some single site. I guess you could use both surf and uzbl for your
day-to-day browsing, but it would take as much time to setup them as
working around the annoyances of opera/firefox/chrome do.

> >Also, there's dillo. Small and fast, but no CSS floats, no javascript.
> 
> the bad part is actually no javascript. since most sites are now
> unusable without.  ( and with, but meh)

Yes, dillo is a curious little bird.


All in all, your attitude seems to be rather harsh towards free software.
But when things start to break one after another, I feel the same. And
I abstain from complaining most of the time, because I remember that the
disappointment is on level with what it used to be when I used commercial
software.

-- 
--  Rogutės Sparnuotos
    this mail is an afternoon's English exercise