new launcher just released

[mouser]

Thanks allen

yksyks -- we should continue this discussion in another thread (maybe here or here -- do you have windows running in large font mode?)

[Tekzel]

FARR's what first introduced me to DC and one of the most used apps on my PC.  Went to use my wife's computer the other day and spent several minutes sure her computer was broken--alt+space did nothing, the catalyst which every application on a computer system relies upon was not present, nothing worked!  She did have something on there called a "start menu", but I am not one for fumbling with ancient technology that may or may not be booby trapped.

-allen (July 19, 2007, 08:34 AM)

Haha, thats funny.  And often the funniest things are true.  I also spend an uncomfortable amount of time (whole seconds!) on a clients computer wondering why "pause/break" doesn't work, only to realize that they of course dont have FARR installed.

[Josh]

I'd love to see the scrollable results window! That would rock!

[mouser]

farr already has this, but perhaps i could add more options about maximum autosize settings to make it more obvious.

[Amadawn]

mouser,

as an old member of donation coder, and someone who used to use FARR all the time and finally dropped it in favor of Launchy I think that I might give a different perspective on why "less featured" programs such as Launchy and Dash appeal to some (perhaps many!) users.

First of all, let me say that I _loved_ FARR. When I discovered it I was blown away and I inmediatly became a donationcoder member and I donated to FARR. I still like it and I think it is useful but I don't use it anymore.

There were 4 main reasons why I started using Launchy instead of FARR:
- "Better" search results (out of the box, without tweaking).
- Looks awesome compared to FARR (although not as good as Dash)
- Some really cool pluggins (e.g. the calculator and web search pluggins)
- Simpler setup and better "out of the box" experience.

Please, before you guys jump to dismiss these reasons, let me elaborate a bit more. But first, please note that the last time I tried FARR it was a few months back and perhaps some of these have been fixed since them. I'd be cool if you pointed me to those.

1. Why I think that the Launchy search is better?
- It is "index-based" which means almost instantaneous search results. There was a noticiable difference compared to FARR.
- "Adaptive" search results. Launchy tracks what I select after every search I do and boosts the ranking of those results the next time I repeate the same _OR A SIMILAR_ search.
- Non contiguous search (i.e. searching for ff will find FireFox)

I believe that some of these features were planned for FARR v2.0 but I don't know if they all made it. I would agree that the "index-based" search may be a drawback for some people, which want to be able to search a completelly up to date list of files, but in my usage I've found that using Launchy with a 20 minute update timer is good enough 99.9% of the time (and when it is not updating the index is 2 mouse clicks away). Perhaps this works well for me because for me Launchy is mainly a program launcher and an interface to google (and a calulator and folder navigator...) and not a file search tool. But I think that there are better "file search tools" around that both FARR and Launchy anyway.

2.- Better look. I think that this cannot be questioned. I don't even think that Launchy looks "that good" but it definitely looks way better FARR. I know that a lot of people don't really care about this but _A LOT_ of people do care (me included).

I think that saying that "power users" do not care about looks is untrue. Actually I find that many power users like to tweak the looks of their computers a lot, as they use them so much more than non power users.

I really don't think that you should under-estimate the importance of this. One of the previous posts made a very good point regarding commercial vs freeware software and I think that there is a reason why commercial software "tends" to look better and that is because most people like better looking software and sometimes are willing to trade some features for good looks and style (case in point ipod vs other mp3 players, etc).

3.- Better "out of the box" experience:
This is related to #1. There is an enourmous amount of options in FARR. I bet that the options pane is probably overwhelming for many users and maybe you should consider create two separate "simple" and "advanced" "configuration modes" to make FARR more approachable and feel less bloated (often bloat is just a feel, it does not need to be real to put some people off).

However that was not what I mean about a better "out of the box" experience and why I stopped using FARR. I know that I can configure FARR to behave exactly as I want with all the very useful and detailed options. I actually did that when I was using it. However, I often go to new computers and I like to have a keyboard launcher in them. With Launchy, I can just install it and it works very well without any configuration. With FARR I would need to waste some time (probably a while!) to configure it properly in each machine that I install it.

A previous poster mentioned that trying to use his wife's PC without FARR was a pain. I bet that trying to use FARR without any reconfiguration would be a (much smaller) pain too!

I really believe that this last reason is very important and probably something that with some careful thought could be easily improved in FARR. You could think about the most useful scenarios for FARR use and make sure that FARR works seamlessly in those cases. Making google searches, opening firefox/explorer bookmarks, finding start menu items, being able to navigate and find the most important system folders (my documents, etc) out of the box should be a priority.

Ah! There is also another (small) criticism to FARR. The name (at some point I almost convinced mouser to change it into something else! :-). Launchy, Dash, SilverLight, QuickSilver... quite frankly, they all sound better than FARR. OK, I know. That is a pretty stupid complaint, but it is true, nevertheless ;-)

I think that most of this can be easily improved and that _technology_ wise FARR is (under the hood) probably the most flexible and impressive launcher for windows. But these loose ends stop it from reaching its potential, IMHO.

Cheers,

Angel

[mouser]

Thanks Angel for some very reasonable comments -- critiques like this will always be welcome on this site.

I'm in the middle of working on something right now so i'll reply more in detail later.  Some of these issues have been implemented with version 2, some have not.

[jgpaiva]

I think i'll give some help, since mouser is busy right now.

#1:
  index: Yep, no index, and you're right as it needs to be done. Still won't be included in V2, though.
  adaptive search: done
  non-contiguous search: not done (i also find it a bit confusing), but multi-word works (f f would find firefox)

#2:
  better look: I've talked with mouser just yesterday about this. It indeed does make a difference for many people, and it'd probably get more people to use farr, BUT, the time it'd take to change the looks would probably be better spent in improving other parts of farr, at the expense of losing some "costumers" because of the looks.
  Also, farr's looks have been dramaticaly improved with the new "large icons display", even though it still isn't as eye-candy as the competitors.
  IMO, even more important than the looks, are the webpage and the screencasts, which make a huge difference on first contact with farr.

#3:
  simple/advanced mode: done.
  default configurations:

making google searches, opening firefox/explorer bookmarks, finding start menu items, being able to navigate and find the most important system folders (my documents, etc) out of the box should be a priority.

yep, all of those work out of the box, although the bookmarks part is done through a plugin, which might not be included in the default package (not sure, as this is mouser's decision).
Another way to solve this problem would be to just carry farr on your usb disk, since farr is fully portable

Also, farr can now work as a calculator, a process killer, a clipboard enhancement, a service starter/stopper, program unninstaller, etc etc etc, all through plugins.

As for the name... I do agree that there could be better names, but i think it's a bit late to be changed.


Thank you for your post, amadawn, it has contributed in 2 ways: to remember what else still needs to be done, and how well farr is now compared to other launchers
Quoting mouser on our conversation (i hope he won't get upset ):

<mouser> it IS a very huge leap from v1
<mouser> in fact when i think on v1 now i get sick

Maybe you should take a look at farr V2 beta, or wait for its grand launch next week. I think you'll be positivelly surprised

[mouser]

At some point we should probably have a discussion about ways to improve the looks of FARR.  As someone who had to be dragged kicking and screaming to add skin support for FARR, i'm a hard sell.. I tend to prefer the standard windows interface.

Having said that, here are some screenshots of FARR demonstrating some of the visual options now available in version 2, including custom font settings and large icon mode:

[Josh]

Mouser,

Have you considered ridding yourself of the titlebar for farr? Thats one thing I like about all of the other launchers I've seen. They are just the search box, with a results box which pops up after entering text. Perhaps you can A. Rid us of the titlebar area, B. add an option to only show previous results if you want them to show, and C. Clean up the options dialog into a more modern and organized fashion. This would make it more aesthetically pleasing for most of your users. I am willing to design the options dialog if you would be interested, just let me know.

Josh

[lanux128]

BUT, the time it'd take to change the looks would probably be better spent in improving other parts of farr, at the expense of losing some "costumers" because of the looks.

jgpaiva was spot-on with this point. before the days of plugins in Farr, i was tempted to use Launchy not because of the looks but because there was a plugin for searching Firefox bookmarks. this plugin feature has been implemented in Farr since then and has expanded to several other useful plugins.. and also that Farr allows me tweak more than other launchers is another appeal, this is my personal opinion, of course..

[mouser]

Have you considered ridding yourself of the titlebar for farr?

look in the above list and you'll see a picture without a titlebar -- it's a skin created in fact by a member here on this forum, rbeeger: http://www.donationc...dex.php?topic=7949.0

[Armando]

In my opinion the only two options  amadawn mentioned that FARR 2 really “lacks” are the index and the uncontiguous search (but launchy lacks many features that Farr has).

So… An index, yes... and Mouser is well aware of that aspect (might even be using locate's?). But I’m sure Farr wouldn't be as great without also keeping its current search mode.  I’ve used quicksilver and Launchy, and, well, when I rename files — which I do all the time — I want my files to be available now. Not in 20 min, or after I manually reindex my HD.

Uncontiguous search? that would be nice, but I wonder if it really adds that much usability. Maybe.

Oh, and there’s the look thing. For me, if it’s functional, I couldn’t care less about the looks. And I actually like FARR’S look. Not very glamour if you use the windows default, but... Compared to other launchers, with farr I can see more files with more details (except for picture view — but then, I prefer to use something like faststone or something like that)… And I need that level of detail since Farr is also my file explorer for quick navigation.

Anyway, Farr is far from ugly -- the slenderFARR.skn is actually quite sexy -- it’s just not… flashy. But yes, a quicksilver bubbly shiny candy look might attract more people. Won't be using though. My windows still has the 95 looks...

[mouser]

we should probably figure out what term we want to use for the kind of non-contiguous/uncontiguous search modes that farr and others use.

Farr v2 now support non-contiguous search.
If you type: f fox OR f f OR fox fire then farr will find firefox, and it will try to be smart about scoring, in that the closer you come to typing the program name with words in the right order, the higher the score, which should help to differentiate near false positives.

What FARR will not do on its own is let you type a string of letters without spaces and match those at different places.  So you can't type: feto find firefox (you could of course very easily with a few clicks add an alias to find any specific program you like with any specific alias keyword).  I'm not sure what that kind of matching should be called (intra-word non-contiguous matching?).  It's possible we could add this one day but i'm reluctant to take the speed hit it would incur.

[Armando]

What FARR will not do on its own is let you type a string of letters without spaces and match those at different places. 

-mouser (July 25, 2007, 12:10 AM)

Yes. I thought that was what uncontiguous meant in that context...

Personally, I don't care to much about "type a string of letters without spaces and match those at different places". I'm happy with "f f" (instead of "ff"). I usually type the first syllable of a word anyway. Faster and I don't have to memorize patterns.

Maybe the performance hit wouldn't be too bad IF there was an index? Dunno...

[mouser]

An index does speed up search.  Sometimes very significantly, such as when searching in huge folders (the start menu search which is what FARR is used for 99% of the time, wouldn't gain you much).

An index does have some drawbacks, as cited above.  If it's not up to date on recent changes it can miss new stuff, and if it is forced to always be up to date it can result in lots of wasted cpu keeping it updated.  It also means more memory use.

FARR will eventually get indexing added as an optional feature.

[Armando]

as an optional feature.

-mouser (July 25, 2007, 12:29 AM)

I'm all for it too... as an option. 

[lanux128]

+1 for indexing as an optional feature..

[nitrix-ud]

I personnaly love the fact that FARR searches without an index

so +1 for indexing BUT as an optional feature

what would be really great is the following :
specify in a "dosearch" command what kind of search you want to use

for a huge folder you want to use the index
for instance :

[Select]

Music album search $$1 | dosearch -alias +folder_music +open_winamp +\ +USEINDEX $$1

i would use it !!

[justice]

It's the fonts that would need tweaking to improve the visual aspect not so much the skin, you can easily have it look like this:

I've attached the .ini file I use so feel free to use the font settings. If you use the whole ini then all your settings will be overwritten so that's probably not recommended.

[Select]

EditEditBoxHeight=28
EditResultLineHeight=20
EditSlideLineHeight=11
EditBoxFont.Name=Garamond Premr Pro Smbd
EditBoxFont.Size=14
EditBoxFont.Pitch=0
EditBoxFont.Style=1
EditBoxFont.Color=-2147483640
SmallResultFont.Name=Garamond
SmallResultFont.Size=12
SmallResultFont.Pitch=0
SmallResultFont.Style=0
SmallResultFont.Color=-2147483640
LargeResultFont.Name=Garamond
LargeResultFont.Size=28
LargeResultFont.Pitch=0
LargeResultFont.Style=0
LargeResultFont.Color=12632256
RicheditFont.Name=Consolas
RicheditFont.Size=10
RicheditFont.Pitch=0
RicheditFont.Style=0
RicheditFont.Color=-2147483640

[ak_]

I really agree with nitrix, we should be able to select indexing (or not) for each alias individually. For example, i wouldn't use indexing to search start menu, but i definitely would to browse my mp3 collection.

[mouser]

we should move discussion of indexing options and methods to another thread..
perhaps the main reason i haven't implemented indexing yet is because there are a lot of decisions that need to be made about the best way to store data and what kinds of options should be provided.

[Amadawn]

mouser (and others),

thanks for your answer and specially for showing off the new skin capabilities in FARR v2. It certainly shows and improvement over FARR v1. It is still not as nice as other launchers but it is better than before, which is great. BTW, is it possible to disable the numbers in the "big icons" mode?

Regarding the "look" of FARR, one thing that I like in Launchy, Dash and Quicksilver is how they really "highlight" the current selection. That is, the element would be launched if you clicked ENTER. I think that this should be prominent in the interface, not just be the first result in the list of search results. I also like how Dash and Quicksilver highlight the letters that are being matched (this is more important when you have intra-word non-contiguous matching, which FARR currently lacks).

As a final note on "eye candy" and "looks". I love the way QuickSilver fades in and out when you use it. That sort of thing could be implemented without modifying the current skin capabilities of FARR. It could even be a plugin, with several fade-in/out animations to chose from (and obviously it shoul be totally optional). Maybe that is an easy way to differentiate FARR in the looks department compared to other launchers?

I will certainly give the new version a try once it is released. I am happy to see that a few of the weaknesses in the previous version have been improved.

One particular comment that I'd like to make is that I think that the "real" intra-word non-contiguous matching (sometimes called "fuzzy matching") is really a _must_. It does not seam a big deal until you start using it but it is really useful and once you get used to it you may never want to use a launcher that does not have it. When coupled with "automatic learning" (i.e. results re-scoring based on user selection) it becomes a really nice way of "automatically" creating keywords.

What I mean is that, in FARR, it is often suggested to create an alias when you want to use a certain name for a certain program or file (say ff for Firefox). But that is the whole point! I _don't_ want to have to manually create keywords. But that means that you have to do that for every computer where you use Launchy (unless you use a usb key or something). And it takes longer than letting the launcher observe what you do and learn what you want it to do. Not to mention that it is also harder for non power users to use this feature.

It is hard to explain, but it really makes a big difference. And having to add a space between the letters would be annoying (the best example is firefox, which I currently can start by simply tapping twice the "f" letter).

And this is another reason for indexing, I think. As you said, implementing this sort of "fuzzy matching" may be too costly in a pure search based program like FARR. But for index-based launchers this is not a problem whatsoever.

I really liked the idea that a previous poster had, were you could select whether you want to index or search each individual folder. I also think that it may be nice if you could do "both" at the same time. That is, quickly present the results of the index search, and then do a search in the background and show any additional results as soon as they are ready. Maybe this could bring the best of both worlds?

Another question that I have is: how does the new version of FARR handle "alternative" modes? I really like how in Launchy if you select "Google" (by typing gg for instance) and type TAB you go into "google search mode". In Dash this is even more evident, as the "Google search" text appears on top of your editbox which makes it clear that any words that you type will be used for a google search.

Another related question is how does FARR handle "alternative actions". By default, hitting ENTER opens or executes the program or file selected. Can you select things such as "open folder" or "copy", etc? In Dash and Quicksilver you can by hitting TAB until you select the action that you want to perform.

As a final note I'd like to answer jgpaiva regarding the trade off between improving the FARR interface versus adding new features. I have to say that the looks of an application can often limit its capabilities. For example, QuickSilver has a really neat "multi-level" "stack" interface, where you can select a file, type tab, select and action, type tab and select a "modifier" for the action (e.g. select a filename, select "copy to..." and select the folder to which the file will be copied). I don't think that this is something that is too difficult to implement. But in Quicksilver it is very natural as every time that you hit tab the selected item is "saved" into a "stack" that is shown in the interface very clearly. So the interface flexibility not only looks good but also adds a good deal of functionality, making it trivial to add this sort of feature. In FARR (and sadly, also in Launchy) this would not be nearly as easy to do, as they are based around the concept of a single edit box where you type your command. That could never be as powerful as the stack based interface approach that QuickSilver (and to an extent Dash and the new SkyLab from Candylabs) has.

Anyway, I'm really looking forward to the FARR v2 release. I don't think that I could give up "fuzzy" matching at this point, but maybe you guys will change my mind? :-)

Cheers,

Angel

[mouser]

BTW, is it possible to disable the numbers in the "big icons" mode?

yes.

can you elaborate on what the "automatic learning" thing is?

Another question that I have is: how does the new version of FARR handle "alternative" modes? I really like how in Launchy if you select "Google" (by typing gg for instance) and type TAB you go into "google search mode". In Dash this is even more evident, as the "Google search" text appears on top of your editbox which makes it clear that any words that you type will be used for a google search.
Another related question is how does FARR handle "alternative actions". By default, hitting ENTER opens or executes the program or file selected. Can you select things such as "open folder" or "copy", etc? In Dash and Quicksilver you can by hitting TAB until you select the action that you want to perform.

FARR is designed to work a little bit differently than these, though i'm considering adding some kind of hybrid modes.

With farr, most of these things like searching the web are done through regular expression aliases.
So for example as you type "search blah" you trigger the search alias and see:


This is a menu of actions to perform on the search text, and as always hit Enter to launch first one, or choose the one you want to launch directly with function/alt+# or click, etc.

Farr also can use +ACTION expressions to say you want to perform some action on an item, for example jgpaiva added an action to zip files and folders:

I guess these are cases where FARR is trying hard to let you type what you want to do as fast and efficiently as possible, sacrificing a multiple-step action with one string specifying the entire command.

one thing that I like in Launchy, Dash and Quicksilver is how they really "highlight" the current selection. That is, the element would be launched if you clicked ENTER. I think that this should be prominent in the interface, not just be the first result in the list of search results.

One of the things that is different about FARR compared to the others is that it really treats the result list as a kind of menu.  It's true that you can hit enter to launch the first result, but while most of these other programs show one results and then after some delay show some extras like a hint, FARR has always been designed to present you with top matches, which is even more important in aliases, where often the results act like a menu.

QuickSilver has a really neat "multi-level" "stack" interface, where you can select a file, type tab, select and action, type tab and select a "modifier" for the action (e.g. select a filename, select "copy to..." and select the folder to which the file will be copied). I don't think that this is something that is too difficult to implement. But in Quicksilver it is very natural as every time that you hit tab the selected item is "saved" into a "stack" that is shown in the interface very clearly.

I think this is something we are going to have to think about, i can definitely see the value of it for some things, and in walking the user step-by-step through a series of choices, i'm just still a bit doubtfull about how often it would really be used vs. the paradigm of finding a file and launching it.  But i'm not at all opposed to implementing a kind of stack-like thing.

Amadawn I hope you will keep trying FARR every once and a while and reporting what it would need to be a viable replacement for what you are currently using.  My guess is that it's going to be a while before you would switch to FARR, but your comments/insights/suggestions can only improve the program, so keep them coming. 

[Amadawn]

mouser,

thanks for taking my comments into consideration :-)

Let me try to explain what I mean by automatic learning. Basicallly, what it means is that everytime that you select a certain search result, the score of the selected search result increases in two ways. First the overal score of that item (file or program) increases for any search that matches its name. Second, the score of that particular item for the set of letters that you used to find that search result also increases greatly.

For instance, let's say that I want to start visual studio. If I typed "visual" it would probably be one of the first (if not the first) result in Launchy. But let's say that I want to start visual studio with "vs" instead. Probably, if I just type "vs" Microsoft Visual Studio will not be the first result to appear in the list. Maybe it will be the 5th or 6th, for instance. I can then use the down key and select it (i.e. I press ENTER). Then, the next time I type "vs" Visual Studio will be the 1st or 2nd item in the list. If it is the 2nd I can just select it again and the next time it will certainly be the 1st result. Launchy automatically learnt that I want to use "vs" as the keyword for Microsoft Visual Studio.

The neat thing is that Launchy also scores higher any previously selected items compared to items in its index that have never been selected. So, imagine that when I first typed "vs" Microsoft Visual Studio were not even on the list. No problem. I can type "visual" that first time (or maybe just "vis") and find and start visual studio. Then the next time I type vs Visual Studio will certainly be on the list as Visual Studio does match "vs" and now it has a higher score.

I don't know the exact heuristics that Launchy uses, but it is really cool. You can feel how Launchy is learning what you want. Very often it will do the right thing right away, but when it doesn't it is very easy to "teach" it. There is no need to use menus and keywords to configure it.

Now, imagine that later on you install some other program called "version control", for instance and for some reason you don't want to use "vs" to start visual studio anymore, but to start this new program. Well, you simply need to select "version control" a few times (once or twice) after typing "vs" and Launchy will quickly learn that you changed your mind. Visual Studio will still be shown on the list of results, probably as the second option.

By the way, Launchy _also_ has a pretty flexible keywords plugin
(installed and enabled by default) but in my experience you rarelly need to use it as the automatic learning, coupled with fuzzy matching works fine in most cases. The only times that making a keyword is necessary is when you want to use a keyword that does not match the file name or if you want to be able to pass parameters to the program that the keyword starts.

BTW, I may be making sound all of this too fancy, with names such as automatic learning and fuzzy matching. The whole point is that the launcher should just pay attention and try to do what I want it to do :-)


Regarding the stack vs the +ACTION modifiers (which by the way I think are pretty clever and cool) I think that the stack based approach has an advantage in that, once you select one item in the stack, the launcher can use its "search" capabilities to match the next element on the stack.

Let me explain with one example:
Imagine that I want to send an email to you and I want to attach a certain log file to the email.

In QuickSilver I can first search for the log file, called "My log.log" for instance. Imagine that I type "log" and find "My log.log". I type TAB. The default action was "open" but I type "e" which finds the "email as attachment" action. I type TAB. I type "mou", which finds "mouser@donationcoder.com" on my address book. I type ENTER. A new email is open in Outlook, addressed to mouser and with the file "my log.log" as an attachment. Total keys typed:

"log[TAB]e[TAB]mou[ENTER]" -> 10 in total.
How many would it take with the +ACTION approach? FARR cannot continue matching after the 1st element is found, can it? And even if it could, it would be a bit ackward, because if it starts showing you the email addresses on the results list, then you lost the name of the file that will be attached (as you did not type the whole filename, you just looked for it my typing a few letters). With the stack based GUI, you keep all the information, and if you make a mistake you can go back easily (just type ESC to go back to the previous stack level).

Also note how after you type each "TAB" the "search context" changes in a Stack based approach. Initially it searches for files and programs. Then it searches for actions and finally, depending on the action type it may search for something else (in this case for email addresses). Plus, it you have automatic learning you can just type "e" for the "email as attachment action" because the launcher learnt that you want to use "e" to select that particular action.

I agree that these types of usages are a bit advanced. But when the interface makes this powerful capability extremely accessible, even for non power users.

So again, looks can improve the program capabilities in some cases. I think that this is a very good example.

BTW, as I said, I really appreciate you guys taking my comments into account. I will definitely keep checking on FARR from time to time.

Angel

[mouser]

FARR does score higher any programs that you have launched in the past, so that it will score things higher if it sees you have chosen them before, but the idea of boosting scores of non-first entries specifically related to the phrase typed is an interesting one.  To my knowledge FARR was the first program of this type to use a variety of heuristics for scoring items and so i'm always interested in new heuristics.. that's certainly an interesting one, and one that wouldn't be very hard to implement.  From a practical standpoint I do wonder if it would really make a difference once you boost the scores of programs you have launched in the past, since that by itself is usually enough to boost any program up to the top.  But then again in cases like your latter your comment about learning that the letter e should be associated with "email attachment" might be a case where it would help.. I'm not entirely sure i want to go down that road of learning though.. I am wary of things that try too hard to learn what you want to do.. I get visions of Clippy the MS clipboard "helper", and i generally prefer things where it isn't that hard to figure out why something got the score it did.

Your comments about TAB changing the search context and allowing filtering to then switch to a different set of options (ie switching between files and actions) is spot on the money, and that's a real benefit of that approach, and why it would be nice to have such a feature.  We've actually discussed it at length in one of the threads here, but i keep putting it off while i cogitate on the best ways to integrate it.