A new video from the SkyOS site shows off the IndexFeeder and the Viewer displaying live results. The video was captured in VMWare showing a few queries made over a 40GB attribute and file content indexed filesystem. There is also a document explaining a few internals of the indexing system used in SkyOS.
SkyOS Indexing
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSAlert.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I continue to be amazed at how fast SkyOS progresses. The IndexFeeder is simply awesome… Great work!
-G
Looks like skyos is using attributes and quering better than beos from which the skyfs came.
The user interface portion of the IndexFeeder that is currently showcased in the video is just the beginning of the IndexFeeder’s capabilities; this will eventually be merged with the design of the new Viewer and Album Viewer seen earlier (see links below).
http://www.skyos.org/images/viewer_mockup_001.png
http://www.skyos.org/images/viewer_mockup_album_viewer_001.png
Module of indexation in the file system, this not always best decision. I am anymore inclined to indexing_server, which was
offered by MAUI-team.
Man, that looks like a really nice desktop! I love the GUI.
http://shots.osdir.com/slideshows/slideshow.php?release=298&slide=1…
SkyOS is cool. If only it was free for personal use, I would download it and try.
instead of archiving media files for image calalogs and music management utilities (e.g. itunes), how about archive the fingerprint instead of the path to the file. this would allow a user to organize their own files, move them around, eliminate duplicates very quickly, and still perform fast pre-cached meta-data information. Plugins would allow applications to request the indexing engine to archive particular data about certain file types/extensions.
I’ve always thought that these utilities should allow the user to set “auto import” directories where you place your media files. move them, no problem. rename them, no problem. as long as they are below an “auto import” directory.
I’d think even a partial checksuming algorithm would have great performance and few conflicts, probabilisticly. The real beauty would be if there was a way to let the indexing engine run on top of any filesystem, or optionally place some hooks in common ones at the driver level so that file moves and renames would be constant time updates to the indexing engine.
just an idea…
Robert Szeleney is like a coding robot, the Axel D"orfler for SkyOS, maybe even better. He’s actualy including ideas from Linux,Apple,Windows,BeOS into a new OS that looks familiar.
If he continues to code at this rate he will surpass both Linux (distros) and Windows pretty soon
He should get in contact with a really good GU-Designer. Nothing in SkyOS follows Fitt’s Law.
I can’t play the video on OS X. Any hints? I tried Quicktime, vlc and mplayer.
I can’t play the video on OS X. Any hints? I tried Quicktime, vlc and mplayer.
http://www.techsmith.com/download/ensharpendecoder.asp
Donwload/install that and the video will work in OSX.
He should get in contact with a really good GU-Designer. Nothing in SkyOS follows Fitt’s Law.
I couldn’t agree more. I’ve tried to bring some sense into the SkyOS community on this, explaining them what could be changed in order to improve, but it was to deaf ears. The team never really responded, the community itself didn’t even know Fitts’ Law (‘It’s just a law, who cares about a law?’), and they wanted to keep everything as Windows as possible, grossly overestimating the power of predisposition.
Don’t try to talk about sensible GUI design in the SkyOS community. You’ll get a ‘SkyOS has themeing capabilities’ (but no docs on how to theme) and that’s it.
@T Shane Bostick:
>> The real beauty would be if there was a way to let
>> the indexing engine run on top of any filesystem,
>> …
The IndexFeeder (a service running in the background) is able to index all filesystems. (SkyFS, BFS, FAT and ISO9660). If a filesystems supports attributes, the IndexFeeder manages two seperate index catalogs, a SQL based content index and an attribute index which is actually part of the filesystem itself.
If a filesystem doesn’t support attributes, the IndexFeeder will store meta data in a seperate SQL based meta data index. From the user view, there is no difference if a filesystem supports attributes or not. You can just search for them.
>> or optionally place some hooks in common ones
>> at the driver level so that file moves and
>> renames would be constant time updates to
>> the indexing engine.
The IndexFeeder automatically gets notified by the kernel whenever a file content changes or a file was created/moved/renamed/deleted. (This works for every filesystem). Because of this the IndexFeeder is always up to date.
@Michael Gaspid:
Only the attribute index is part of the filesystem (which btw has quite a few advantages). The content index and an optional attribute index for filesystems not supporting attributes is handled by the IndexFeeder service.
The IndexFeeder service is also responsible for handling queries. The multiple thread architecture of the IndexFeeder makes it possible to query filesystems, search the SQL database and update the widget showing the results at the same time.
@Thom Holwerda:
>> I’ve tried to bring some sense into the SkyOS
>> community on this, explaining them what could
>> be changed in order to improve, but it was
>> to deaf ears
Thom, discussing what could be done and what is bad and what should be changed doesn’t really help. Comming up with words like “Fitts Law” doesn’t help either. The are various threads about the SkyOS GUI. But designing a new GUI from the information in this threads, will not really work.
What would help would be a mockup, a sample implementation, a concept, a specification, which respects Fitts Law.
Interested in making one? (You know, this is much more time consuming then “just” talking about a new GUI concept).
And please don’t condemn people about not knowing what Fitt’s Law is. They don’t have to know.
The developers have to know about it. And once there is enough time for redesigning various GUI parts, this will be done. But currently this is not on the TODO list or in the near roadmap.
>> the community itself didn’t even know Fitts’ Law
>> (‘It’s just a law, who cares about a law?’),
>> and they wanted to keep everything as
>> Windows as possible, grossly overestimating
>> the power of predisposition.
You really should stop listening to 12 year old people. Or at least, you should know, who represents the SkyOS development and who not. You know what I mean.
Thom, please try to stop this discussion here on osnews. If you want to talk about this further, you know how you can contact me.
Well I’m sure everyone at SkyOS community knows this mockup, but some shamefull self promotion is always nice ^_^
Here -> http://www.deviantart.com/view/14519620/
A really nice mockup done by me, but somehow I feel it won’t be seeing the light… (get it?)
What would help would be a mockup, a sample implementation, a concept, a specification, which respects Fitts Law. Interested in making one? (You know, this is much more time consuming then “just” talking about a new GUI concept).”
Robert, I’ve done a mockup. It’s in those threads, as you will probably know. Also, I made a guide on how to exactly mimick my ideas in KDE. That’s even better than a mockup. You know, even SkyOS’ lead graphics designer made & posted a mockup!
The mockup + explanation by me:
http://www.skyos.org/board/viewtopic.php?t=19611&start=0&postdays=0…
My guide as to how to achieve it on KDE:
http://www.skyos.org/board/viewtopic.php?p=77423#77423
SkyOS’ lead graphics designer mockup, based on my ideas:
http://www.skyos.org/board/viewtopic.php?t=19715&start=0
Those are just a small selection of threads. In total, there are about 7 threads discussing this issue. You couldn’t have missed the ones above, Robert.
And please don’t condemn people about not knowing what Fitt’s Law is. They don’t have to know. The developers have to know about it.
Yeah, I know people don’t need to know Fitts’ Law; my point is: if you know fcuk all about something, then don’t talk about it. The kids in the SkyOS community don’t really get that, that’s obvious.
But hey, what am I to do? There were enough people inside the community willing to make the GUI subject of a proper discussion (heck, each member of the SkyOS team asked me to bring this up in the forums multiple times!); yet, the kids came in and flamed the discussions dead. I want something like this to be discussed in public. That is impossible in the SkyOS forums and impossible on irc as you don’t go there anymore (understandable). I don’t want to discuss this over IM; two people cannot discuss a UI, that must be done with a group of sensible people.
Thom, please try to stop this discussion here on osnews. If you want to talk about this further, you know how you can contact me.
As said above, discussions concerning UI should be made in public. If this requires heavy moderation on the forums to weed out the kiddos, than so be it.
Ditto…
I see you guys are aiming at a more OS X like interface. Problem is it’s easy to make gradients and alpha stuff, but to do it right and make it usable is another thing! Why don’t you contact a/some skinner(s) from thee Neowin forum? There are 2 or 3 people who are really good and I bet they are interested!
And what I think would be a great idea is to implement 2 interfaces. One basic (without flashy gradient stuff) and one shiny, BUT in the same spirit. Not like Win9x and Luna look for example but something coherent. I know there are a lot of guys out there who don’t like shiny stuff.
I’ll probably sign up for the beta testing soon, I like the way you’re heading! Keep it up!
Hella, its ugly, but read carefully, it’s not about looks
.
Yes, I know this threads. The problem is that this threads are just various bitmaps and texts spread all over the forum. (Nothing one can work with)
One would have to bring all the important information, mockups, and constructive comments from all this threads into a nice OO/PDF document.
This way we are able to see all the important stuff at once and have one common place to edit, update and talk about the designs.
We need something like a system/functional specification for the entire GUI design topic.
I definitely agree. I’d be more than willing to bring all this together and bring some structure in it, in pdf or something– however, posting it on the forums would be useless.
I sense it’s time for a Wiki.
“kids” hahaha, that’s rich coming from you, guy. real rich.
thom i hate to break it to you but the vast majority of your “gui discussions” have been you wanting a close button on the left, a minimize button on the left and double click maximize and then you calling any one who disagreed an idiot.
for the most part that was it, a lot of bitter comments on a small aspect of the gui, maybe in IRC it’s different but i’ve never seen you post on say, the moving start button that exists a pixel or two below the top of the screen lessening the advantage that putting things on and edge gives you.
people keep giving you “it has themes” because shit like window controls are traditionally (in the *nix world) done according to whatever the theme tells the WM to do and as such is a pretty minimal gui concern.
I see that Sky-OS is looking good but is it useable and how do you get it?
(Useable meaning hardware support etc?)
follow the link in the article, and all your questions will be answerd.
and i must say that sky os looks to be one of the more interesting developments in the os world right now. in that they do something similar to beos, take a step back and start from scratch.
i dont wonder tho, what type of kernel do they use sky os? ie, is it similar to how the linux kernel is designed or is it more like a micro of nano kernel?
SkyOS uses its own kernel, built from the ground up.
I had heard of SkyOS before, but never really paid attention to it. I decided to give the website a look around. Needless to say, I was more than a tad bit disappointed to find it to be a closed source project, especially after apparently being an open source project prior to 5.0.
And it’s not so much about having to pay to use it – I think $30 dollars is more than a reasonable amount to pay for an operating system.
Linux has taught me that being able to audit and modify the source code of an operating system to suit your needs is a Good Thing(TM). It seems that they’re a very developer friendly community, but not being open source is a major turnoff.
Just my random and anonymous 2 cents.
I still can’t believe this guy keeps adding new features to the OS, instead of making a stable “ready for sale” release.
I see, every time a new “popular” feature comes for any os, this guy implements some stuff alike for SkyOS.
I foresee indigo, avalon, webcore, audiocore and all the new stuff around implemented in SkyOS prior to a stable (non-beta) release.
Just my thoughts …
that do still not explain what kind of design they have used on the kernel. ie, will i be able to unload a driver and load a new one without having to restart the whole system?
That I would not mind, heck of a lot cheerier and with more “spirit”, so to say.
“I had heard of SkyOS before, but never really paid attention to it. I decided to give the website a look around. Needless to say, I was more than a tad bit disappointed to find it to be a closed source project, especially after apparently being an open source project prior to 5.0. ”
Actually it was FREE before 5.0. It hasn’t been Open since before version 3, and that was actually a completely different OS anyway, just basically the same name.
And I don’t get what’s wrong with Closed Source! Open source has way more problems than closed. If it’s closed, everything works with the OS. If it’s open, a “Linux” application designed for redhat might not work on SuSe.
I commend SkyOS for choosing to keep the source closed.
99% of the time if the compiler used is the same it will work on many distros. what you will run into tho is diffrent locations for stuff like config files and installation tools.
hell, for a long time there have been a alien tool thats able to transform a rpm (used by red hat, suse and others) into a deb (used by debian and derivatives) and it should work nicely.
point is that all these package formats are basicly renamed tar-balls with some added files to tell the database and tools where the diffrent files are supposed to go and what to do if your updating a old version (overwrite, with or without backup, the configfiles or leave them alone and write the new one next to it so that the admin can look it over and see what have changed to adapt the old config into the new file).
only a few times have binary compatiblity shown up and then mostly in red hat as they tryed to get ahead of the pack by using a development version of gcc or similar.
allso, isnt skyos using a lot of open source software, but recompiled to run on their kernel? they would have had a lot less apps this early in development if they could not help them selfs to the open source apps out there.
Yes, I had read that prior versions were open source/free. And I agree, there are benefits to closed source operating systems, including the assurance that “software will just work”.
And it’s not that there’s anything wrong with the software. It’s the fact you can’t audit the code, or modify it to meet your specific needs, or learn from it!
Windows is a good example of this. There’s tons of great software available for it. Microsoft offers an expansive API for developers to utilize. But, unfortunately, it’s closed source. You can’t audit the code and you have to “trust” that Microsoft does The Right Thing. I certainly wouldn’t equate SkyOS with Windows when it comes to trust, but its still disappointing to see a project in this day and age closed to the public.
But don’t get me wrong – I will heartily agree that one of Linux’s greatest strengths – its openess – is also its greatest weakness. Being pulled in TOO many directions can weaken (or at least slow) the development of even great software. Sometimes one unambiguous direction is the best path for development, but I’d rather see that direction chosen by a community than a board of directors or a very select few.