Default Firefox settings

[Arthur Pemberton]

Hello,

I recently asked about ways to benchmark Firefox[1] because of how
slow I found Firefox to be  on Fedora 7. It was almost unusably slow,
prompting me to switch to Konqueror as much as possible.

However I have found a short tutorial which after following the speed
seems to at least be on par which my laptop (which has half the
processing speed, and a slower, IDE hard drive, and slower memory)

Assuming I haven't encountered some freak situation, it may be
necessary to consider changing the default Firefox settings, one of
which is IPV6 DNS. I have zero issues against IPV6 myself, but if it
helps make Firefox useful, by all means.

Again, if I can provide any objective metrics, please let me know.

[1] http://www.redhat.com/archives/fedora-desktop-list/2007-September/msg0013...
[2]
http://www.zolved.com/synapse/view_content/28106/How_to_speed_up_Mozilla_...

-- 
Fedora 7 : sipping some of that moonshine
( www.pembo13.com )

[Jeroen van Meeuwen]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Arthur Pemberton wrote:
>  Hello,
>  
>  I recently asked about ways to benchmark Firefox[1] because of how
>  slow I found Firefox to be  on Fedora 7. It was almost unusably slow,
>  prompting me to switch to Konqueror as much as possible.
>  
>  However I have found a short tutorial which after following the speed
>  seems to at least be on par which my laptop (which has half the
>  processing speed, and a slower, IDE hard drive, and slower memory)
>  
>  Assuming I haven't encountered some freak situation, it may be
>  necessary to consider changing the default Firefox settings, one of
>  which is IPV6 DNS. I have zero issues against IPV6 myself, but if it
>  helps make Firefox useful, by all means.
>   
Does upstream think the default settings as provided in the vanilla
distribution should be adjusted as set forth in [2] ? If they don't, I
think we shouldn't either.

>  Again, if I can provide any objective metrics, please let me know.
>  
>  [1] http://www.redhat.com/archives/fedora-desktop-list/2007-September/msg0013...
>  [2]
> http://www.zolved.com/synapse/view_content/28106/How_to_speed_up_Mozilla_...
>   
Kind regards,

Jeroen van Meeuwen
- -kanarip
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.7 (GNU/Linux)
>
> iD8DBQFHA8u4KN6f2pNCvwgRAgYiAJ929YYrJhgeTrjaatAvNdJdsQISZwCg0cmW
> 5ZomvUCCswashuY/N66qEuI=
> =Bxa1
> -----END PGP SIGNATURE-----

[Christopher Aillon]

Jeroen van Meeuwen wrote:
>  Does upstream think the default settings as provided in the vanilla
>  distribution should be adjusted as set forth in [2] ? If they don't, I
>  think we shouldn't either. 
Several of the preference tweaks will indeed enhance performance for a 
reasonably large percentage of websites but will cause a not 
insignificant number to break.  (By refusing to load, hanging the 
browser, etc.)  The prefs are generally that way for a good reason.

[Dennis Jacobfeuerborn]

Christopher Aillon wrote:
>  Jeroen van Meeuwen wrote:
> > Does upstream think the default settings as provided in the vanilla
> > distribution should be adjusted as set forth in [2] ? If they don't, I
> > think we shouldn't either.
>  
>  Several of the preference tweaks will indeed enhance performance for a 
>  reasonably large percentage of websites but will cause a not 
>  insignificant number to break.  (By refusing to load, hanging the 
>  browser, etc.)  The prefs are generally that way for a good reason.
>   
Does there exist any useful statistical data out there to verify the "not 
insignificant number" claim? I've been browsing the web with pipelining 
enabled for ages and have yet to come across a page that doesn't work. Is 
the problem real or one of these myths that stay around because people read 
on the internet that pipelining has problems so it must be true?

Regards,
   Dennis

[Christopher Aillon]

Dennis Jacobfeuerborn wrote:
>  Does there exist any useful statistical data out there to verify the
>
>  "not insignificant number" claim? I've been browsing the web with 
>  pipelining enabled for ages and have yet to come across a page that 
>  doesn't work. Is the problem real or one of these myths that stay around 
>  because people read on the internet that pipelining has problems so it 
>  must be true? 
You could look to see how many Apache 1.3.x webservers are out there 
still via Netcraft, etc.  Unless it's had a patch to it (I've not kept 
up on Apache 1.3).  My blog post [1] should help you figure out what 
data to look for if you want to figure this out.

[1] http://christopher.aillon.org/blog/dev/mozilla/20050105-pipelining.html

[Christopher Aillon]

I wrote:
>  You could look to see how many Apache 1.3.x webservers are out there
>
>  still via Netcraft, etc.  Unless it's had a patch to it (I've not kept 
>  up on Apache 1.3).  My blog post [1] should help you figure out what 
>  data to look for if you want to figure this out. 
Also, see
https://bugzilla.mozilla.org/buglist.cgi?quicksearch=sw%3A%5Bpipelining%5D 
which has a bug list of things that need fixing before enabling 
pipelining by default, including some sites which break.

[Arthur Pemberton]

On 10/3/07, Jeroen van Meeuwen <kanarip(a)kanarip.com&gt; wrote:
>  -----BEGIN PGP SIGNED MESSAGE-----
>  Hash: SHA1
>
>  Arthur Pemberton wrote:
>  > Hello,
>  >
>  > I recently asked about ways to benchmark Firefox[1] because of how
>  > slow I found Firefox to be  on Fedora 7. It was almost unusably slow,
>  > prompting me to switch to Konqueror as much as possible.
>  >
>  > However I have found a short tutorial which after following the speed
>  > seems to at least be on par which my laptop (which has half the
>  > processing speed, and a slower, IDE hard drive, and slower memory)
>  >
>  > Assuming I haven't encountered some freak situation, it may be
>  > necessary to consider changing the default Firefox settings, one of
>  > which is IPV6 DNS. I have zero issues against IPV6 myself, but if it
>  > helps make Firefox useful, by all means.
>  >
>
>  Does upstream think the default settings as provided in the vanilla
>  distribution should be adjusted as set forth in [2] ? If they don't, I
>  think we shouldn't either.
>
>  > Again, if I can provide any objective metrics, please let me know.
>  >
>  > [1] http://www.redhat.com/archives/fedora-desktop-list/2007-September/msg0013...
>  > [2]
> http://www.zolved.com/synapse/view_content/28106/How_to_speed_up_Mozilla_...
>  >
>
>  Kind regards,
>
>  Jeroen van Meeuwen
>  - -kanarip 

Fair enough, but the speed differences cannot be ignored. Nor is the
fact that it makes Firefox (at least on Fedora) to seem very slow.


-- 
Fedora 7 : sipping some of that moonshine
( www.pembo13.com )

[Chuck Anderson]

On Wed, Oct 03, 2007 at 01:15:53PM -0500, Arthur Pemberton wrote:
>  > Does upstream think the default settings as provided in the
> vanilla
>  > distribution should be adjusted as set forth in [2] ? If they don't, I
>  > think we shouldn't either.
>  >
>  > > Again, if I can provide any objective metrics, please let me know.
>  > >
>  > > [1]
> http://www.redhat.com/archives/fedora-desktop-list/2007-September/msg0013...
>  > > [2]
> http://www.zolved.com/synapse/view_content/28106/How_to_speed_up_Mozilla_...
>  
>  Fair enough, but the speed differences cannot be ignored. Nor is the
>  fact that it makes Firefox (at least on Fedora) to seem very slow. 
Actually, I use IPv6 here and it isn't the cause of Firefox slowness.  
My biggest problem is simply rendering speed.  It is excruciatingly 
slow in scrolling an already loaded web page.

On the bright side, for whatever reason the F8 version of Firefox 
seems much faster.  Can anyone explain what might have changed?

[darrell pfeifer]

I'm using firefox granparadisio.

I was having a terrible time with unbearably slow firefox on some
hotel networks. All my other network apps were working fine.

It looked like firefox was timing out on dns lookups. Eventually when
addresses got cached locally on the machine firefox speeded up
considerably.

In preferences/network/setting, "auto detect proxy settings" was set.
I changed to "direct connection to internet" and firefox works at a
decent speed.

darrell

[Arthur Pemberton]

On 10/3/07, darrell pfeifer <darrellpf(a)gmail.com&gt; wrote:
>  I'm using firefox granparadisio.
>
>  I was having a terrible time with unbearably slow firefox on some
>  hotel networks. All my other network apps were working fine.
>
>  It looked like firefox was timing out on dns lookups. Eventually when
>  addresses got cached locally on the machine firefox speeded up
>  considerably.
>
>  In preferences/network/setting, "auto detect proxy settings" was set.
>  I changed to "direct connection to internet" and firefox works at a
>  decent speed.
>
>  darrell 

When firefox is slow, it is _really_ slow. Is this acceptable by default?


-- 
Fedora 7 : sipping some of that moonshine
( www.pembo13.com )

[Nicolas Mailhot]

Le Jeu 4 octobre 2007 01:03, Arthur Pemberton a écrit :
>  On 10/3/07, darrell pfeifer <darrellpf(a)gmail.com&gt; wrote:
> > In preferences/network/setting, "auto detect proxy
> settings" was
> > set.
> > I changed to "direct connection to internet" and firefox works at a
> > decent speed. 
>  When firefox is slow, it is _really_ slow. Is this acceptable by
>  default? 
You do need to auto detect proxy settings - when a proxy is available
most of the times you won't be allowed direct internet connection.

The problem is firefox sucks big time when you're behind a proxy - as
soon as the proxy is slow to answer for one tab all the others start
being sluggish/frozen (I suppose firefox has internal locking where
the cost of some locks is only evident in a proxy context)

The problem is not the setting, the problem is firefox is buggy.
That's something to report upstream.

-- 
Nicolas Mailhot

[Bernie Innocenti]

Nicolas Mailhot wrote:

>  You do need to auto detect proxy settings - when a proxy is
> available
>  most of the times you won't be allowed direct internet connection.
>  
>  The problem is firefox sucks big time when you're behind a proxy - as
>  soon as the proxy is slow to answer for one tab all the others start
>  being sluggish/frozen (I suppose firefox has internal locking where
>  the cost of some locks is only evident in a proxy context)
>  
>  The problem is not the setting, the problem is firefox is buggy.
>  That's something to report upstream. 
I experienced this problem to.  My guess was that Firefox blocked while
resolving hostnames in proxy.pac JavaScript snippets like this one:

 function FindProxyForURL(url, host)
 {
  ip = dnsResolve(host);
  if (isInNet(myIpAddress(), "10.0.0.0", "255.0.0.0")
    && !shExpMatch(url, "ftp://*")
    && !isInNet(ip, "127.0.0.0", "255.0.0.0")
    && !isInNet(ip, "10.0.0.0", "255.0.0.0")
    && !isInNet(ip, "81.174.33.43", "255.255.255.248"))
   return "PROXY 10.3.3.1:3128";
  else
   return "DIRECT";
 }

-- 
 \___/
 |___|  Bernardo Innocenti - http://www.codewiz.org/
  \___\ One Laptop Per Child - http://www.laptop.org/

[Matthew Miller]

On Thu, Oct 04, 2007 at 10:05:07AM +0200, Nicolas Mailhot wrote:
>  The problem is firefox sucks big time when you're behind a proxy
> - as
>  soon as the proxy is slow to answer for one tab all the others start
>  being sluggish/frozen (I suppose firefox has internal locking where
>  the cost of some locks is only evident in a proxy context) 
I think this is because network.http.max-connections-per-server, which
defaults to 8, applies to the *proxy server* when using a proxy. Crank that
number way up (as much as whoever is running your proxy will allow) and
watch that problem go away).




-- 
Matthew Miller           mattdm(a)mattdm.org          <http://mattdm.org/>
Boston University Linux      ------>              <http://linux.bu.edu/>

[Nicolas Mailhot]

Le Jeu 4 octobre 2007 14:43, Matthew Miller a écrit :
>  On Thu, Oct 04, 2007 at 10:05:07AM +0200, Nicolas Mailhot wrote:
> > The problem is firefox sucks big time when you're behind a proxy -
> > as
> > soon as the proxy is slow to answer for one tab all the others start
> > being sluggish/frozen (I suppose firefox has internal locking where
> > the cost of some locks is only evident in a proxy context)
>
>  I think this is because network.http.max-connections-per-server, which
>  defaults to 8, applies to the *proxy server* when using a proxy. Crank
>  that
>  number way up (as much as whoever is running your proxy will allow)
>  and watch that problem go away). 
Unfortunately this does not seem to work. For some obscure reason as
soon as a page contains stuff hosted at a site the proxy does not
like, all the firefox UI freezes waiting for the proxy to relay it (it
won't), instead of  recovering gracefully and processing the other
stuff (like it does on direct connexion when a website is dead)

-- 
Nicolas Mailhot

[Arthur Pemberton]

On 10/4/07, Nicolas Mailhot <nicolas.mailhot(a)laposte.net&gt; wrote:
>  Le Jeu 4 octobre 2007 01:03, Arthur Pemberton a écrit :
>  > On 10/3/07, darrell pfeifer <darrellpf(a)gmail.com&gt; wrote:
>
>  >> In preferences/network/setting, "auto detect proxy settings" was
>  >> set.
>  >> I changed to "direct connection to internet" and firefox works at a
>  >> decent speed.
>
>  > When firefox is slow, it is _really_ slow. Is this acceptable by
>  > default?
>
>  You do need to auto detect proxy settings - when a proxy is available
>  most of the times you won't be allowed direct internet connection. 
I don't think I turned of auto proxy detect to get the speed boost.

-- 
Fedora 7 : sipping some of that moonshine
( www.pembo13.com )