[Kst] 2.0.7 first regression testing + ideas for the future

[Nicolas Brisset]

Hi,

so, I have been playing around with the current svn version and here are my observations. I'm writing down everything that comes to my mind, it does not mean everything has to be addressed before releasing 2.0.7! Some points should probably go to bugzilla so that we can keep track of them.
1) when changing the ASCII configuration options (particularly field names) the comboboxes further down the dialog are not updated automatically. Workaround: apply the settings by clicking OK, and reopen the config options. It would be nicer to have the comboboxes update automatically if it does not slow down the interface too much
2) when saving vectors to disk and they have different lengths I thought there was a dialog asking what to do (interpolate, truncate, etc). It no longer seems to be the case, though I haven't checked exactly what happens, though I guess it interpolates. Why not, but an option to ask the user may be better, sometimes interpolation is not wise.
3) moving objects (plots or annotation objects) can't be undone
4) in the view strings dialog all objects appear, even those without strings (equations, plugins, etc...)
5) adding a curve can't be undone with "Undo"
6) the view vector dialog is very inconvenient to compare vectors (we desperately need a table view) and also because the precision/formatting of the displayed values can't be set
7) we haven't concluded on the discussion where to add "quick plugins" like differentiate or cumulative sum, which we'd like to have in the RMB menu but not under "Filter". My proposal would be to create a new category... "Quick plugins"
8) Sometimes I can't create a shared x-axis box, I wonder if it's not linked with the z-order of the plots. When dragging down and one is not above the previous one, or some have been moved around, it sometimes does not work. 
9) thinking about the recent suggestion to be able to shift curves around with the mouse (provided by someone using some other tool), we could actually do the following: extend the shift plugin to shift in x or y (by taking 2 input vectors and 2 input scalars) and then map for instance a middle mouse button click + drag to a move of an image of the curve (as plot dragging currently) and upon MMB release it would compute the deltas to apply and create the new curve using this plugin. Sounds cool, eh?
10) Another cool idea (it's late, I'm getting creative :-)): add an action to the Range category to restrict the vectors to currently visible parts. I think I mentioned already in the past, but I'm not sure. It would allow to compute statistics on subparts of the data, as was discussed on the list at Kst 1.x times.
11) sometimes I get dotted curves even though they should be plain lines (seems to be when width = 0)
12) when loading big files, some feedback in the status bar would be nice. And the cursor also shouldn't revert to normal (I looked at the code and thought it would be easy to fix, but I can't really figure out why the standard cursor is restored too early)
13) automatic scalars could need a small cleanup: some names are not straightforward + some scalars are missing
14) we haven't concluded on the discussion I actually failed to initialize to the appropriate location for the new ASCII options (file buffer limitation + buffer size + threads). As I said before, those options depend more on the PC being used than on the file itself, so maybe a global setting would be more appropriate?
15) I think the "New Session" action does not clean up everything as it should. If I look at memory usage, it confirms that: I have a big session loaded with 450 Mb RAM usage, click on New Session and... still have that RAM usage. Something is not clean there. Plus I have always had the feeling that using this new session led to instability. Unfortunately, I don't have a reproduceable scenario
16) I fear there has been a regression in the ASCII reader very recently. It seems to work under Windows (though I see huge performance variations but there's dfrgntfs.exe running...) but under Linux I can't get the data to load, which is strange as it worked before all the recent changes! I always get "The file could not be read because not enough memory is available". Strange...
17) I got one crash: ASSERT: "isPtrValid()" in file /home/nicolas/Code/kst_clean/src/libkst/sharedptr.h, line 170

Well, that's it for now. I think this is really going to be a solid release. The above list may look impressive, but I more or less went through all the menus and tried a lot of the features and found only those issues (or ideas for future improvements). 
If I can vote, I'd say we have to fix before the release the following points: 4, 8, 14 now or never, 15 if possible, and 16.

Nicolas