


System Software by arkascha 20 comments

Dolphin Service Menus by oliverthered 2 comments
If one doesn't exist it creates one in ~/.kactivities
the activity name is the directory root name.
so for instance if I have ~/Projects/Clients/FooBar I could start and activity there.
or ~/Videos I could start one there...
etc....
This is just a proof of concept (files are only a couple of lines long... read readme)
- Apr 01 2011

Utilities by marcomaniac 232 comments
I've pulled the latest source and will put up a PPA at some point as I couldn't see one. - Feb 18 2011

Various KDE 1.-4. Improvements by Linuster 15 comments
I haven't downloaded this re-incarnation of the original 3.x version I wrote many years ago before going off to the DirectX 9 3D for Wine and then having some health issues. (well medication issues!)
anyhow,
The two things I'll add to this extension are, the ability to drop a file into clipboard:/ and have it copied to the clipboard contents and as someone mentioned in a prior post, integration with kliper (though I'm not sure how easy that's going to be, it may be an idea to abstract out the history into kio_clipboard and then integrate kio_clipboard into kliper instead of the other way around.
I suppose then the ability to archive off large portions of the clipboard, or move them or delete entries or add multiple entries (effectively creating a history by dropping more than one file onto it)
Now to find out if my other extensions / modifications made it into 4.x or not! - Sep 01 2010

Various KDE 1.-4. Improvements by oliverthered 26 comments
To test with KDE I have done the following:
installed KDE 3.4.1-r1 as usual,
then
emerge kde-libs
and CTRL+c after ./configure has run
do the same with kde-base
then apply the patch to kde libs and kde base in /var/tmp/portage/kde-libs-4.3.1-r1 and /var/tmp/portage/kde-base-4.3.1-r1
then
make install kio in kde-libs
make install kcontrol in kde-base
make install libkonq in kde-base
There was a problem with a missing header (I forgot to note which one), just cd into the directory and
make nameofmissingheader.h
then start
make install .. again. - Jun 20 2005

Various KDE 1.-4. Improvements by oliverthered 26 comments
I haven't had any crashes though and kcontrol starts fine for me.
I used kde-base 3.4.1-r1 to test the patches.
I've also just find a crash problem, with view in view only, clicking refresh causes a crash. I'll try to fix this soon. - Jun 20 2005

Various KDE 1.-4. Improvements by oliverthered 26 comments

Various KDE 1.-4. Improvements by oliverthered 26 comments
There's also an option to load only the icons in view because currently if you fire up 3 or 4 konqueror file views all those view will attempt to load all their icons at the same time, hogging CPU and memory resources. - Jun 18 2005

Various KDE 1.-4. Improvements by oliverthered 26 comments
The first was caused by comparing a non-pointer to NULL, the second was caused by kdebase needing kdelibs to be installed first. - Jun 18 2005

Various KDE 1.-4. Improvements by oliverthered 26 comments
but this is easly fixed.
either use a cast to (char char *)
or string = new QString("this is a char *")
I'll find out what's causing the second problem as soon as my system updates finished. - Jun 18 2005

Various KDE 1.-4. Improvements by oliverthered 26 comments
I'm recompiling kdevelop at the moment so it may be a little while. - Jun 18 2005

Science by virtualmeet 138 comments
Is it not possible to have a 4d object and then transform the functions be a view vector to create a 3d projection.
Other methods you can use are root solving (the povray source is a good place to start looking).
You find one point that has a non-imaganiry solution, then you start walking unit boxes, projecting a ray and seeing if you have a solution along each edge of the box.
this is mainly useful for polynomials but can be used for other objects such as blobs. - Jun 17 2005

Various KDE 1.-4. Improvements by oliverthered 26 comments
But that's not how konqueror works at the moment. With or without the patch the same amount of time is spent loading thumbnails (well unless you selected to load only those in view).
All this patch does is change the order in which the thumbnails are loaded, putting those icons in view at the top of the list.
If you change the view port say by scrolling then a moment afterwards the icons will be re-orded so that those in view are loaded first.
Because it takes a moment, the order will not be updated until you stop (or pause) you scrolling. - Jun 17 2005

Various KDE 1.-4. Improvements by oliverthered 20 comments
view / preview
called Executables
you just need to make sure it's ticked and previews are turned on...
If it's not there you need to copy pethumbnail.desktop from src to
KDEROOT/share/services/
If you still having problems you need to make sure that kde has the mime type application/windows-executable associated with *.exe, *.dll and *.ocx files.
To check this use the menus to goto
settings / configure konqueror / File Associations.
Type exe in Find filename pattern, and
application -> windows-executable should be listed...
if it isn't then you need to copy
windows-executable.desktop
from src to
KDEROOT/share/mimelnk/application/
Let me know if your still having problems...
I'll check that the installers doing it job. - Jun 17 2005

Various KDE 1.-4. Improvements by oliverthered 5 comments
Max would be very obvious, min would only highlight operations that are tightly linked to the selected object. - Jun 17 2005

Various KDE 1.-4. Improvements by oliverthered 26 comments
I have some other plans to sort out browsing but there going to take a while to implement...
Things like if a page hasn't been viewed for an hour put it into cache to free up some system memory. - Jun 17 2005

Various KDE 1.-4. Improvements by oliverthered 5 comments

Various KDE 1.-4. Improvements by oliverthered 20 comments
here's the bug to get this into kde.
http://bugs.kde.org/show_bug.cgi?id=107491 - Jun 15 2005

Science by virtualmeet 138 comments
You can use either animation as if taking slices through an object, or shadow techniques where the shadow has three dimensions instead of two.
This would be really cool for helping people visualise object with more than three dimensions. - Jun 15 2005
I've found FileLight quite useful but very slow. If you want to replace KDiskFree I would suggest that you performance and memory tune FileLight, here are some suggestions that could increase performance and reduce memory all are reasoned all are costed (if the cost isn't the reason). If you don't know how to implement or are unsure of any of the algorithms then drop me message. Some of my suggestions my already be partly implemented via the 'cache' mechanism, if so they do not appear to be working for me.
all notes are against 1.0beta_6.
'make install
Making install in src
make[1]: Entering directory `/var/tmp/portage/filelight-1.0_beta6/work/filelight-1.0-beta6/src'
Make tree nodes as light as possible, if states only apply to a few nodes remove the states from the node and keep a separate list. (vtables partly count here so make sure you using static and non-virtual functions etc.. as appropriate)
Reasoning: The more of the tree that fits in memory,cpu cache the better. and there's no point storing needless information in nodes. Locating the state is a n for the number of states, not n for the number of nodes. Memory overhead is per state + 1 not per-node.
(Does Chain really need to be a double linked list? for every directory?)
Start by using a queue system to the next node, then a stack system when you get a number of nodes deep, This way you can generate the chart as you traverse the directory tree.
Reasoning: a queue will cause all directories in the current directory to be scanned next, a stack will causes the first then second descendants to be scanned, since the user wants to know about the 'current' node, all children of the current node should be scanned before any descendants. Using a queue will probably increase memory usage when compared to a stack.
(cost slight memory overhead + instant visual display vs reduced memory overhead, delayed visual display)
Make a simple version with no directory traversal for those people who just want a df, without hammering their server.
Reasoning: I just want a df, I don't want to wait ten minutes while my 1tb of data gets scanned turning my server into a lemon.
(cost a few inodes almost instant vs every inode on my fileing system)
When you traverse the nodes stack style consider aggregating the results.
Reasoning: unless a file some way down the file hierarchy is especially large there is no point in showing it so there is no point in storing a node for that file, just aggregate with the parent directory. (cost fixed memory vs n memory)
Use a weighted balanced tree to decide when to aggregate. (or calculations based on a weighted balanced tree).
Reasoning: this enables you to aggregate in an intelligent manner based on filesize, tree depth, tree width etc....
(cost is balanced tree with aggregation (=log2n ~= logn) as apposed to no aggregation n)
Deletes should run off of cache, or don't update the tree for a delete.
Reasoning: I'm deleting files, not updating the tree, let me delete in piece. A lazy update could run in the background, so long as it doesn't affect the user deleting. (cost is 1 delete vs 1 delete + n inode operations)
Only parse the aggregate node and it's children when a user "cd's" to the aggregate node or it's parent, don't recalculate the whole tree.
Reasoning: The user is obviously interested in the current node, even if the other nodes may be a few meg bigger/smaller don't bother updating unless the user asks. (maybe show an out-of-date Icon or something) (cost is 0, except an icon, vs n inode operations)
Don't delete parent nodes (from the internal tree) until you hit a memory limit, that way hitting back is a as near zero time operation (unless you've hit the memory limit, in which case sections of the tree will need reparsing).
Reasoning: The use may want to hit the back button (it is the most commonly used button apparently!), so don't clean out cache that could be used to generate an almost instant back. (cost is 1 memory look-up, and a fixed memory cost, vs n inode operations)
Make tree nodes as light as possible, if states only apply to a few nodes remove the states from the node and keep a separate list.
Reasoning, locating the state is a n for the number of states, not n for the number of nodes.
Memory overhead is per state + 1 not per-node.
Start by using a queue system to the next node, then a stack system when you get a number of nodes deep, This way you can generate the chart as you traverse the directory tree.
Reasoning: a queue will cause all directories in the current directory to be scanned next, a stack will causes the first then second descendants to be scanned, since the user wants to know about the 'current' node, all children of the current node should be scanned before any descendants. Using a queue will probably increase memory usage when compared to a stack.
(cost slight memory overhead + instant visual display vs reduced memory overhead, delayed visual display)
Make a simple version with no directory traversal for those people who just want a df, without hammering their server.
Reasoning: I just want a df, I don't want to wait ten minutes while my 1tb of data gets scanned turning my server into a lemon.
(cost a few inodes almost instant vs every inode on my fileing system)
When you traverse the nodes stack style consider aggregating the results.
Reasoning: unless a file some way down the file hierarchy is especially large there is no point in showing it so there is no point in storing a node for that file, just aggregate with the parent directory. (cost fixed memory vs n memory)
Use a weighted balanced tree to decide when to aggregate. (or calculations based on a weighted balanced tree).
Reasoning: this enables you to aggregate in an intelligent manner based on filesize, tree depth, tree width etc....
(cost is balanced tree with aggregation (=log2n ~= logn) as apposed to no aggregation n)
Deletes should run off of cache, or don't update the tree for a delete.
Reasoning: I'm deleting files, not updating the tree, let me delete in piece. A lazy update could run in the background, so long as it doesn't affect the user deleting. (cost is 1 delete vs 1 delete + n inode operations)
Only parse the aggregate node and it's children when a user "cd's" to the aggregate node or it's parent, don't recalculate the whole tree.
Reasoning: The user is obviously interested in the current node, even if the other nodes may be a few meg bigger/smaller don't bother updating unless the user asks. (maybe show an out-of-date Icon or something) (cost is 0, except an icon, vs n inode operations)
Don't delete parent nodes (from the internal tree) until you hit a memory limit, that way hitting back is a as near zero time operation (unless you've hit the memory limit, in which case sections of the tree will need reparsing).
Reasoning: The use may want to hit the back button (it is the most commonly used button apparently!), so don't clean out cache that could be used to generate an almost instant back. (cost is 1 memory look-up, and a fixed memory cost, vs n inode operations)
Only update the counter at the bottom right had side of the screen for every... (suggestion: directory entered or exited).
Reasoning: Writing text is surprisingly CPU intensive, the use can also read a value that changes 10times a second, but can't read one that changed 100times a second.(Cost, Moderately reduced CPU overhead, improved clarity reduced granularity vs improved granularity, greater CPU overhead, reduced clarity).
Two column view doesn't fit on my screen,the number of columns must be dynamic or one e.g. two lots of ...'
/net/bigdisk/films on /var/data/fake_windows/MySharedFolder/movies type none (rw,bind)' side by side won't fit 1280x1024.
Columns should be left (or right for some languages) aligned not centred.
Reason: I have to allign the columns in my head if they are centred, but not if they are left/right aligned. headings centred, lists aligned.
you MUST use getmntent to read the fstab if you want to replace KDiskFree, see 'man fstab' or http://www.rt.com/man/getmntent.3.html for more info.
Reason: fstab says, 'The proper way to read records from fstab is to use the routines getmntent(3)',
what if fstab is in digieridoo format?
I hope you implement some of my recommendations, they should help make your application business class.
Reason: What's good for the goose is good for the gander, why not make the best possibly application. - Feb 26 2005
it's much simpler than the code you currently have, and it meens that your application should work on and system that impments getmntent regardless of the format or location of fstab
man fstab notes....
The proper way to read records from fstab is to use the routines getmntent. - Feb 21 2005

Various KDE 1.-4. Improvements by IceSheep 15 comments
You filter would then be a white list instead of a blacklist, which is much better from a lock-down point of view. - Feb 19 2005

Various KDE 1.-4. Improvements by IceSheep 15 comments
1: A user will always demand that they need what they already have, change is a hard thing for them.
2: When you get use to the change you'll probably not want to go back to the old way, because it's another change.
Even you would benefit from some files being hidden e.g. KDE hides .* file by default, turn them on and you get lots of noise, turn them off and you can still access them if you know where to look. - Feb 19 2005

Dolphin Service Menus by elchile 13 comments

Various Stuff by jekko 7 comments
It's still a little buggy but is maturing quite quickly. - Feb 03 2005

Various KDE 1.-4. Improvements by lokheed 128 comments
Again most styles seem to come from the 80's when we only had 16 colours and everything had to be a straight line. - Dec 29 2004

KDE Plasma Screenshots by caustiq 10 comments
I'm hoping by some fit women who broke into KDE so they could take it home and keep it just for themselfs. - Dec 22 2004

Various KDE 1.-4. Improvements by boskicinek 112 comments
I think KDE needs a good UI abstraction so that interface problems and functionality problems can be addressed separately.
It would also allow half of the kde 'improvements' to be implemented and tested. - Dec 15 2004

Browser by infernal-quack 8 comments
I was planning at using it to compleatly remove the back/forward buttons from the browser, maybe even the address bar could go if enough people supported doc-links.
Anyhow, I've got enough time to look at getting the plugin working again and check over the implementation but I can't garentee any time in the furture for support and updates. - Dec 15 2004

Various KDE 1.-4. Improvements by WedDa 30 comments
The first is to request a location, and the other is to request a search for a location.
Granted /etc/passwords shouldn't kick off google,
but www.kdeapps.org or trout fishing should.
I would say that the search bar is unintuitive clutter, it's just that you've got a pre-programmed method of using the location bar, so it seems hard for you to think of it as a 'request' bar. I think you would be more 'productive' if the location bar performed searched from day one. - Dec 14 2004

Various KDE 1.-4. Improvements by PaT 10 comments

Developers Apps by amantia 72 comments
Ideal or tabbed MDI with the ability to have a side by side split view.
It's much easier that switching to one tab, making a note, switching back and hopefully putting in the correct data.
A good example would be when translating a page from french to english you need to have both versions in view. - Dec 14 2004

Various KDE 1.-4. Improvements by andrunko 313 comments
since ATI have a crap XRender implementation (and I assume RenderAccel is needed to speed up XRender), I would like to know what the perfomance is like with an ATI card. - Dec 14 2004

Various KDE 1.-4. Improvements by andrunko 313 comments
Some of the ideas look good to me, but some are the kind of thing I would stay clear of, this patch really needs some configuration ability(like how can I turn it off) before it goes anywhere near kde-core.
Swishy curvy transparent things are fine for a 2.5Ghz Athlon 64 with 1gb of ram and a decent graphics card,
they would choke on the 400MHz 64mg ipaq though! - Dec 14 2004

Audio Apps by garycramblitt 6 comments
Usability's high on my agenda at the moment.
I'm looking at tracking how the user uses the UI with the possibility of getting the computer to 'guess' what the user is going to do next and present a cleaner UI to them.
e.g. i have spelling mistakes in this textarea, and when I next press the RMB in this text area I'll probably want to correct them.
Kind of a slightly more advanced
version of Microsoft's hiding menus.
This will be usefully in speech controlled User interfaces, and may be useful in spoken interfaces in determining if something should be read or not. - Dec 14 2004
you can use http://web.tiscali.it/mupuxeddu/software/pitchtrack/ pitchtrack to work out the pitch.
and eric schierer has done some good work on beat tracking.
http://web.media.mit.edu/~eds/ - Dec 14 2004

Various KDE 1.-4. Improvements by oliverthered 26 comments
As you say things like this tend to be good or bad so let me know and I'll hard code the functionality, until then I'll leave it optional and you can try both versions.
If you don't want the option to be available in the GUI don't patch kde-base/kcontrol. - Dec 14 2004

Audio Apps by garycramblitt 6 comments
GCC comment: Borland C would have handed the cast ok, GCC didn't, GCC tends to be a lot more fragile than other compilers.
3ds files, I've tried some fairly old 3ds files (3ds for dos so 6 years + ) , some mot so old ones (from 3dsmax V1) and some new ones From boson. I couldn't get any of them to load. - Dec 12 2004

Various KDE 1.-4. Improvements by oliverthered 16 comments
e.g.
The Querty isn't exactly the easiest keyboard to use, but it's the 'defacto standard' so that's what you get.
Windows is the 'defacto standard' OS and IE is the 'defacto standard' browser, it doesn't make them good, or even standards complient.
2:/
It's clutter, basicly why have to input boxes when you only need one?
I don't like thinks in the top right hand side of my view, I find them distracting, since I'm usually looking to the left of the page and there just out of my field of view. It's a bit like a TV in a pub, flickering away.
Being in the top right, draws my attention away from the work I am doing, and I also have to 'think' about it. e.g. I ocassionally type a search into the location bar, only to have to copy it into the google bar.
I do like it's functionality, well I would if is supported locate:// etc...,
I woudln't get ridd so much as make it redundant, I suppose you can keep it their if you want to, or have the option of removing it but still having intergraed search in kde.
3:/
It would be impossible to the computer to guess exactly what you are thinking ,but it could have a good go.
But without requireing much intelegence from the computer you could bind a seperate shortcut for search as apposed to open, you could get a prompt if there's uncertainty, or you could rediret you focus to the top right of the window and pick a search method. - Dec 10 2004

Various KDE 1.-4. Improvements by arminstraub 178 comments
I haven't started work on it yet, but the idea is to be a bit 'intelegent' about which search to use
say google if your at an http:// address locate if your at a file:// address etc.. and also to pass along some usefull infomation to the search engine, like the current url. - Dec 10 2004

Various KDE 1.-4. Improvements by arminstraub 178 comments
try using kpackage.
If you building from source it shouldn't be too hard to find out what's been installed without using locate.
also try 'find -mtime -1' etc.. to find files that have been reciently modified, accessed or whatever. - Dec 10 2004
Ok, I've tried importing some 3ds files from povray, boson and a very old (dos copy ) of 3dstudio, but get a segment fault (did you know there's a 3ds library available from http://lib3ds.sourceforge.net/
)
Sorry, couldn't get any further. - Dec 04 2004

Various KDE 1.-4. Improvements by oliverthered 16 comments
A miss-typed omain name should try to locate that domain name in google, a miss-typed file name should attempt to find the file, and a non-descript location would be a request to search for whatever you've typed in context.
Anyhow, some people seem to like the idea, and others don't so I'll look at doing an implementation where you can optionally hide the search bar.
I've also got 15-20 people of various computer skill (from none to product testes) who I can buy a pint and get them to do some usability testing, and about 4-5 people who would do long term usability testing. e.g. force them selfs to use a feature for a couple of weeks and see if it is easier touse once your used to it. - Nov 27 2004
I only made my suggestions to help produce a more polished final product.
I'll even do some coding for you if you like so you can see if it's better or worse, - Nov 26 2004

Various KDE 1.-4. Improvements by peppelorum 12 comments
Also if you turn on multiple services there's another horrible black line between the service panes. - Nov 26 2004
Ok,
I've got a good half inch in the middle of the wheel, good enough for 20 charters or so.
The directories aren't in alphabetical, I think it was just the directory I was cleaning out earlier. What order are they in?
I don't know of another app that uses the Directory selection dialogue in KDE which is the dialogue that filelight is using, this makes is seem a bit odd.
And I would expect the screen layout to be the same as Konqueror.
i.e.
the label Location:before the location input.
The back and forward buttons tool-bar the location input tool-bar.
Nice idea, but it feels a little bit odd to use at the moment. - Nov 25 2004

Various KDE 1.-4. Improvements by maarizwan 13 comments
If I click the rmb on the page I get a waking 13 options to choose from,
Back, forwards and relaod are at the top, and I never use them. If they were hidden I would only have 10 options to choose from.
I never use set-encoding, stop-animations or copy to either.
So that leaves 7, almost half the original number of entries.
The same thing goes for most of the menus, i've never used window/move-tab left so I should be shown the option.
Hiding things I don't use improves my perfomnace, customizes the UI to my needs automaticaly and makes it less probably that I will click on something that I don't want. - Nov 25 2004