Ubiquity 0.5 slow?

4 views
Skip to first unread message

Bhagwad

unread,
Jul 10, 2009, 5:04:17 PM7/10/09
to ubiquity-firefox
Since updating Ubiquity to 0.5, the time taken for the process to
recognize the first few letters, highlight the command and deliver a
local result (such as the "calculate" command) has slowed down
tremendously.

Whereas initially it was milliseconds, it's now at least 2 seconds
long. Anyone else facing this problem? I'm using Ubuntu with Firefox/
Shiretoko 3.5 . Let me boot into Windows and see if the same problem
occurs.

Timur Izhbulatov

unread,
Jul 11, 2009, 5:24:18 AM7/11/09
to ubiquity...@googlegroups.com
2009/7/11 Bhagwad <bha...@gmail.com>:

Yes, I noticed, too, that 0.5 is slower when it's processing my input.
I use Shiretoko 3.5 on Ubuntu 9.04 and have better timing, much less
than 2 seconds, but it's still noticeable, since I have and older
version at work for comparison.

--
Timur Izhbulatov — www.timka.org

"mitcho (Michael 芳貴 Erlewine)"

unread,
Jul 11, 2009, 10:12:26 AM7/11/09
to ubiquity...@googlegroups.com
Some of the new features in Parser 2, particularly the new smart
argument-first suggestions, necessitates that the code be more complex
than in previous Ubiquity releases. There are still some areas I
believe we can continue optimizing the code, however, so this work
will be ongoing.

In particular, Bhagwad, I currently believe we do have a bug where
things take a long time on initial parses (like your first letter
ones), which I'd like to look into.

Thanks for the feedback!

mitcho
--
mitcho (Michael 芳貴 Erlewine)
mit...@mitcho.com
http://mitcho.com/
linguist, coder, teacher

Bhagwad

unread,
Jul 11, 2009, 11:16:25 AM7/11/09
to ubiquity-firefox
Any way I can help out? Need me to do some tests? Just ask

Bhagwad

On Jul 11, 9:12 am, "mitcho (Michael 芳貴 Erlewine)" <mit...@mitcho.com>
wrote:
> Some of the new features in Parser 2, particularly the new smart  
> argument-first suggestions, necessitates that the code be more complex  
> than in previous Ubiquity releases. There are still some areas I  
> believe we can continue optimizing the code, however, so this work  
> will be ongoing.
>
> In particular, Bhagwad, I currently believe we do have a bug where  
> things take a long time on initial parses (like your first letter  
> ones), which I'd like to look into.
>
> Thanks for the feedback!
>
> mitcho
>
>
>
>
>
> > 2009/7/11 Bhagwad <bhag...@gmail.com>:

Bhagwad

unread,
Jul 12, 2009, 7:44:59 PM7/12/09
to ubiquity-firefox
I've turned off the "Use Parser 2 (the next generation parser)"
setting in the Ubiquity preferences page and now the speed is back to
normal.

xero

unread,
Jul 14, 2009, 9:28:12 AM7/14/09
to ubiquity-firefox
I had to turn off Parser 2 in 0.5 as well. Typing causes the entire
Ubiquity dialog to flash on almost every letter. Sometimes it gets
locked up and doesn't recognize that I typed anything, but of course
only in the middle of what I was typing. The parser occasionally just
decides that the command no longer matches when typing its arguments.
I'm hoping the bugs and performance issues get worked out because I
love Ubiquity, but right now it's not even usable.

xero

unread,
Jul 14, 2009, 4:12:59 PM7/14/09
to ubiquity-firefox
It's still flashing on every keystroke and occasionally locking up for
a few seconds even with the new parser turned off. It did that in the
latest alpha as well, but not to the extent it does it now.

"mitcho (Michael 芳貴 Erlewine)"

unread,
Jul 14, 2009, 6:47:19 PM7/14/09
to ubiquity...@googlegroups.com
We're actively working on these issues for 0.5.1. If you get Ubiquity
from the hg repository [1] you can already try out some changes which
may help both the performance and flickering you mention.

[1] https://wiki.mozilla.org/Labs/Ubiquity/Ubiquity_0.1_Development_Tutorial
#Getting_Set_Up

mitcho, Labs

mitcho (Michael 芳貴 Erlewine)

unread,
Jul 15, 2009, 9:17:14 PM7/15/09
to ubiquity-firefox
We've pushed a new 0.5.1 beta which should improve general parser
performance as well as help with the flickering issue. You can get
this beta here:

http://ubiquity.mozilla.com/xpi/ubiquity-latest-beta.xpi

(Note: for techinical reasons it's called 0.5pre7.)

I would love to know if this helps improves the behavior.

mitcho, Labs

On Jul 14, 4:47 pm, "mitcho (Michael 芳貴 Erlewine)" <mit...@mitcho.com>

Bhagwad

unread,
Jul 16, 2009, 8:54:53 AM7/16/09
to ubiquity-firefox
Installed the latest version - looks great as of now even after
enabling parser 2

Will let you know if any issues pop up - great work!

On Jul 15, 8:17 pm, mitcho (Michael 芳貴 Erlewine) <mit...@mitcho.com>

Willy

unread,
Jul 16, 2009, 3:21:46 PM7/16/09
to ubiquity-firefox
Mitcho I must ask why couldn't it have been named 0.5.1beta?

On Jul 15, 7:17 pm, mitcho (Michael 芳貴 Erlewine) <mit...@mitcho.com>

"mitcho (Michael 芳貴 Erlewine)"

unread,
Jul 16, 2009, 3:43:38 PM7/16/09
to ubiquity...@googlegroups.com
We weren't sure whether it would get picked up properly by the update
mechanism as a beta after 0.5... we're pretty sure that's working
right now, so if we release another beta (which I think we may, later
today) we'll call it something like 0.5.1pre2.

m

Sudhir

unread,
Jul 16, 2009, 5:31:49 PM7/16/09
to ubiquity-firefox
I updated to 0.5.1 beta but the issue of slow response remains. It's
taking a while after inputting the first few characters. Even clearing
the commands already typed takes a bit of time. Another thing that I
have noticed is that the time lag occurs mainly when inputting
something that's not a command. This may be something to do with the
parser.

On Jul 17, 12:43 am, "mitcho (Michael 芳貴 Erlewine)"

Willy

unread,
Jul 16, 2009, 5:53:51 PM7/16/09
to ubiquity-firefox
Reason enough to push another beta then.

"mitcho (Michael 芳貴 Erlewine)"

unread,
Jul 16, 2009, 7:27:40 PM7/16/09
to ubiquity...@googlegroups.com
We just pushed 0.5.1pre2 and am planning to push 0.5.1 tomorrow.

(Note 0.5.1pre2's suggestion list flickers/moves around a lot, but
this has already been fixed for the final 0.5.1.)

The perceived "slowness" is actually a combination of a few different
phenomena... one of the major ones (the actual parser performance +
nountype detection tracking, to be exact) has been markedly improved
in 0.5.1... another (the speed of canceling a query when changing
input) is something we hope to attack for 0.5.2.

mitcho
Reply all
Reply to author
Forward
0 new messages