Recent Posts

451

Find And Run Robot / Re: Fantastic app - idea for faster searching

« Last post by nontroppo on December 17, 2006, 09:19 AM »

But FARR still drives lots of disk access - current version causes about ~1800 I/O operations when loading the history before searching anything (I assume mostly reading icon info). Interestingly if I then type in something "pr", reset process monitor to count only new file I/O, then delete my search to go back the history list, FARR performs about ~1800 I/O operations again. It doesn't seem to cache what it pulled up originally on first load - it has to go to disk to refetch the icon data. In this case Farr should cache the data in memory so it doesn't need to go to disk.

452

General Software Discussion / Re: Adobe Flash engine goes Open Source

« Last post by nontroppo on November 23, 2006, 02:59 PM »

For those curious, here are a set of benchmarks profiling the current flash 9 virtual machine against browsers and previous flash versions (artificial benchmarks):

http://oddhammer.com...iptperformance/set4/

There are huge performance gains over previous flash versions, and some substantial improvements over current browser JS engines, but not everywhere. Mozilla/Firefox, as XUL is driven by javascript, can gain some UI performance increases hopefully.

However, it is unlikely other browser vendors will use this engine. I know that opera at least, has to tune their engine for small devices and so use different optimality criteria (memory constraints are sritical etc). Nevertheless, this means more engineers working on standards which is a good thing. Lets hope it will push microsoft to finally update IE to use proper ECMAScript+DOM...

453

General Software Discussion / Re: Firefox Extensions of the Day: Link Alert and TargetAlert

« Last post by nontroppo on November 22, 2006, 01:13 PM »

For opera users, here is the equivalent extension:

http://userjs.org/sc...ancements/link-alert

454

General Software Discussion / Re: Windows memory-paging behavior

« Last post by nontroppo on November 21, 2006, 07:40 AM »

Helped me understand a bit more about windows mem handling (reads easier than most other things I've seen as a non-programmer): http://shsc.info/WindowsMemoryManagement

455

General Software Discussion / Re: IE7 Rant

« Last post by nontroppo on November 10, 2006, 06:16 AM »

tomos: that bug is fixed   - the only existing issue is that the ad-block filter is shared among profiles by default - you need to manually edit the path until that is fixed...

456

General Software Discussion / Re: IE7 Rant

« Last post by nontroppo on November 09, 2006, 10:50 AM »

IE7 has certainly not placated a huge swathe of web developers. Fixes for bugs that took 6 years to fix were touted as substantial improvements. Javascript manipulation of the DOM is still a horrendous mess. Performance on the Acid2 test (advanced style features) looks like opera 4's rendering from the year 2000! Developers were still happy to have crumbs thrown at them, but there are still very good grounds to be upset with it for its pathetic support for modern standards.

If you care about elegant, semantically transparent and interoperable code/information, that is a good enough reason to avoid IE7 for the much better alternative browsers.


Offtopic comments/corrections   

I have my own dislikes/annoyances with Opera - cant create profiles, cant save single window sessions

-tomos (November 06, 2006, 11:45 AM)

for profiles: http://operawiki.info/operaprofiles

to get single window sessions, one would need to manually edit your session file - thus it is possible but not easy (and would be nice feature for the future...) It would be possible for someone to write a third-party "extension" for this though in the meantime...

While Opera is fast to display HTML, and quick loading, the way it handles javascript is very slow. This is a common criticism of Opera, and probably the major reason why I don't use it more than I do.

-Renegade (November 06, 2006, 11:21 PM)

This is absolutely untrue since about V6. For most of ECMAScript and DHTML, opera's current engine is lightning fast. Benchmarks are always incomplete, but here are a quite comprehensive sample of tests that check a whole variety of areas of javascript (opera version is latest weekly, Firefox is the latest trunk build aka FF3):

http://andrew.hedges...eed_test/index3.html

Opera: 0.7s IE7: 3.1s FFTrunk: 3.6s


http://celtickane.co...projects/jsspeed.php

Opera: 0.98s IE7: 6.1s FFtrunk: 3.2s


http://www.24fun.com...benchjs/benchjs.html

Opera: 8.4s IE7: 17.5s FFtrunk: 21.1s (due to popup test mostly)


http://www.speich.ne...er/moztesting/3d.htm - large box

Opera: 5.2s  IE7: 7.2s  FFTrunk: 12.5s

DOM performance can be staggeringly faster than IE or FF - which is very important for well written modern JS:

http://amix.dk/blog/viewEntry/?id=161 - http://nontroppo.org...mer/DOM-compare.html

Basic DOM - opera: 1.1s  IE7: 7.2s  FFtrunk:2.6s


http://www.quirksmod...g/dom/innerhtml.html :

DOM1 - opera: 0.2s IE7: 3.2s FFTrunk: 0.95s
DOM2 - opera: 0.1s IE7: 3.0s FFTrunk: 0.70s

Even IE's very own innerHTML maipulation is faster in opera: 31ms vs. 170ms for IE7


http://www.quirksmod...dom/classchange.html -

class: opera: 62ms IE7: 844ms FFTrunk: 479ms
style: opera: 484ms IE7: 1043ms FFTrunk: 879ms


Opera may perform slower in some places - one of its string concatenation functions is slightly slower (~1.2) than IE and Firefox - this is because opera is optimised for low memory use and deliberately trades that against speed in this specific case (confirmed directly with the developer of the JS engine). Nevertheless, with corner cases aside, opera's javascript performance is very fast; Opera needs to run in devices such as PDAs and even mobile phones after all...

457

General Software Discussion / Re: Sophos no whiner

« Last post by nontroppo on October 31, 2006, 05:02 AM »

Sophos anti-virus does do heuristic/behavioural analysis:

Proactively protect against known and unknown threats

Behavioral Genotype® protection identifies malicious code and blocks it before execution, giving the benefits of a Host Intrusion Prevention System (HIPS).

We have university-wide site licences for it, along with FSecure. It seems pretty efficient and low impact, and fast to scan. Much better than FSecure and closeer to NOD32 in terms of efficiency...

458

TapTap / Re: TapTap - LATEST VERSION INFO THREAD - v1.01.01 - Oct 29, 2006

« Last post by nontroppo on October 31, 2006, 04:52 AM »

yes, this would be useful only if one could define double taps to do one thing and triple taps to do another, thanks anyway mouser! 

459

Find And Run Robot / Re: Idea: launching of farr by pressing the control button twice

« Last post by nontroppo on October 30, 2006, 11:05 AM »

Great stuff!!! :beer:

Is there any plan to roll the double-tap functionality into FARR directly (being the only thing I'd use taptap for)?

460

TapTap / Re: TapTap - LATEST VERSION INFO THREAD - v1.01.01 - Oct 29, 2006

« Last post by nontroppo on October 30, 2006, 11:04 AM »

hey mouser, great app! any chance to specify tap number:

D = double
T = triple
...
C = centuple

etc...

461

Find And Run Robot / Re: Idea: launching of farr by pressing the control button twice

« Last post by nontroppo on October 27, 2006, 04:53 AM »

+1 

462

Find And Run Robot / Re: Startup Speed

« Last post by nontroppo on October 27, 2006, 04:52 AM »

If you don't show history at start, then FARR should be fast to start (it makes almost no file access, everything is in memory). I can only get it to miss MAX one character when I turn history off...

463

Find And Run Robot / Re: Startup Speed

« Last post by nontroppo on October 11, 2006, 05:20 AM »

nitrix-ud - how many search folders - are any network drives? do you start showing history?

also, do you have file monitor from sysinternals[1] - if so how many operations is FARR performing on start?


----
[1] http://www.sysintern...ilities/Filemon.html

464

Find And Run Robot / Re: v1.13.01 test

« Last post by nontroppo on October 09, 2006, 06:20 AM »

Stress testing it (CTRL+ENTER then typing as fast as possible to try to get it to lose the first key press) is promising, only the fastest keypresses can get it to lose characters.

It still seems to use the same number of disk I/O operations on first start with history (~1,300) but is now using about half the number during search (~10,000 vs. ~20,000 before).

There still seems to be a lot of repetitive reading of desktop.ini during search, maybe some of that can be further optimised (store desktop.ini in memory or something)?

BTW mouser, there is a (slightly flaky) PHP script available which parses filemon.log files to give you the top 20 operations performed.

465

Find And Run Robot / Performance Update V1.11

« Last post by nontroppo on October 04, 2006, 05:31 PM »

OK, FARR is now both faster and more useable! I assume that is the memory retention. FARR now sits around 10MB whereas before it would drop to 3MB then go up on activation.

Starting up FARR (I let my history show on a blank search) still causes about ~1400 disk I/O operations. This is multiple reads of each history item - I assume this is reading the icon. It would be good if this could be cached somehow too.

The disk is still showing lots of I/O during search, but FARR is snappier than before.

This is promising indeed mouser!

 

466

Find And Run Robot / Re: comment on new version 1.10.02 beta

« Last post by nontroppo on September 27, 2006, 08:25 AM »

in my heart i've always felt the operating system should be smart enough to cache this stuff, so this is my attempt to see if i can "covince it" to do so.

-mouser (September 26, 2006, 07:37 AM)

well, at least on my system, this is not working. it was some time since i last profiled FARR thoroughly so whether anything here is a regression I don't know.

467

Find And Run Robot / Re: V2 request for fixed launches

« Last post by nontroppo on September 26, 2006, 07:10 AM »

cool feature! 

468

Find And Run Robot / Re: comment on new version 1.10.02 beta

« Last post by nontroppo on September 26, 2006, 07:09 AM »

Hi mouser,

thanks for a new version!  I will run it as my default launcher to help fix any bugs / make improvements...

Could you explain a little at how the new cache works? There appears to be some increase in speed somehow, but there is still a lot of disk access.

I get ~1351 I/O operations on first trigger without entering any text, of which the majority are accesses to desktop.ini - this seems to be FARR getting information on each item in the history list.

I am searching for "opera" - I have only 2 directories, Ian\Start Menu (149 files including .ini's etc) All Users\Start Menu (214 files incl. ini's etc) being searched.

Typing in the first letter "o" gives ~21,000 I/O operations - some >4000 of those are to repeated reads of desktop.ini again. Another keypress "p" causes an additional ~16,000 I/O operations. Deleting a letter "p" again causes another set of ~20,000 operations, it does not seem to have cached "o" typed alone.

I have enabled the experimental cache in preferences. My version say 1.10.05.

IINM the cache is not operational in this situation?

469

Find And Run Robot / Re: Apologies if this is OT...

« Last post by nontroppo on September 11, 2006, 10:39 AM »

mouser: :beer: and :strawberries: for any fixes provided

amadawn: Functionally, launchy is inferior to FARR IMO. For example, the heuristics to determine order in the list are primitive and really non-optimal. I cannot weight entries and of course there are no aliases for power launching. Its only benefit is that, because it is primitive BUT uses a cache, it is quick for the number of shortcuts I have.

Aesthetically, launchy supports alpha-transparency in its skins and so it is going to look sexier. Of course I'd love alpha-transparency in FARRs skin, but that depends on the engine mouser uses. Still, I think FARR looks smart (the skins are nice and the UI flexible) and this is a very minor issue for me.

470

Find And Run Robot / Re: how to use alias / cmd lock pc

« Last post by nontroppo on September 04, 2006, 06:27 AM »

a) what exactly happens when I use : lock PC | nircmd.exe lockws. I am a bit too scared to test it not knowing if I can unlock it again.

-beethoven (August 19, 2006, 11:44 PM)

Don't be scared! All it does is go to the login screen where you have to type your password — it is a standard windows thing nothing custom. You can get nircmd to ask a question first; for example:

[Select]

reboot | c:\windows\nircmd.exe qboxcom "Do you want to lock your workstation?" "question" lockws

b) assuming that this allows me to lock my pc, how to combine this with my screensaver command and where I would have to set up the lock and unlock commands.

-beethoven (August 19, 2006, 11:44 PM)

You don't need both, just get nircmd to run your screensaver (after configuring it to ask password) and the screensaver will ask for your password, so:

[Select]

screensaver | c:\windows\nircmd.exe screensaver

is all you need...

471

Find And Run Robot / Re: Apologies if this is OT...

« Last post by nontroppo on September 04, 2006, 06:13 AM »

Hmm, RAM usage is low here. And, sadly, I've switched over to using it. FARR, as much as it is superior functionally, is just too slow for me; not infrequently I hit the hot key, start typing and lose a couple of letters thus have to backspace and retype, defeating the point of a quick launcher. FARR indexes on-the-fly which simply doesn't scale to the number of shortcuts (not lots, I don't index docs, just program shortcuts) I index with the speed I type. And it rescans on each letter pressed. Part of the problem may be flushing RAM while it is minimised.

Launchy is pretty but dumb — not ideal; but it uses only ~1-2mb more than FARR and comes up instantly. It also doesn't suffer from a long-standing problem with FARR in which alt+space can cause focus issues (I find alt+space ergonomically optimal personally). The focus issue and performance are my current two issues which tips the balance away from FARR for the moment. I'm patiently hoping for a new version one day to fix those.

Mouser — nevertheless thank you for your time and passion and talent, it is important to emphasise that! 

472

Find And Run Robot / Re: "Custom File Explorer" option...

« Last post by nontroppo on August 21, 2006, 08:57 AM »

Salamander doesn't even get called up for me, even though I set salamander, I still get explorer when using "explore here" in the context menu for example...

473

General Software Discussion / Re: Opera and Gmail users might like this

« Last post by nontroppo on July 05, 2006, 12:52 PM »

Well, for me the most annoying thing is that I cannot use advaced mode and hit ´back´ becaue it gets stuck on "loading". I use gmail in standard html mode... or I have to hit reload after back everytime. Do you have that problem?

-urlwolf (June 25, 2006, 02:02 PM)

Is that with V9 - works fine for me.

For more details (and note you can even tweak this per site tuning performance against compatibility using site preferences):

http://www.opera.com...search.dml?index=827

Mark0 - i reported this as a bug already

474

General Software Discussion / Re: Firefox sucking CPU

« Last post by nontroppo on July 05, 2006, 12:48 PM »

I'm runing the latest firefox 1.5.0.4.  I noticed some lag in my system and found that Firefox is sucking up 50% of the CPU. 

-Rover (June 21, 2006, 11:13 PM)

Try removing all your extensions - the firefox core shouldn't do that. extensions, being 3rd-party, can often have memory and other resource leaks as they are not so extensively tested. you can then add them back one-by-one until you find your culprit.

475

Find And Run Robot / Re: Progress?

« Last post by nontroppo on July 02, 2006, 09:35 AM »

plea for FARR 2: