a bunch of weird bugs

[Thomas Friedrichsmeier]

Hi,

On Wed, 14 Sep 2016 19:51:24 +0200
meik michalke <meik.michalke at uni-duesseldorf.de> wrote:
> Am Mittwoch, 14. September 2016, 18:09:49 CEST schrieb Thomas
> Friedrichsmeier:
> > On Wed, 14 Sep 2016 12:51:36 +02000
> > meik michalke <meik.michalke at uni-duesseldorf.de> wrote:
> > If you monitor the output file in dolphin while running a plugin,
> > do you see the timestamp updating?
> 
> yes, the file is being updated immediately (and plots are being
> added).

No good idea on this, immediately, then. Please check the
rkward.frontend log. Problems watching the file for changes _should_
show up, there, hopefully along with something helpful.
 
> btw, in the virtual machine the import/export file dialogs show the
> same buggy behavior in the kf5 build as i've recorded here:
>  http://reaktanz.de/stuff/R/screencast_RKWard_menu_bug.avi
> 
> i.e., they are simply unusable.

Hm. This would be easier to pin down, if I could see it, too. The one
unusual thing we do with those menus is me make the same menu
accessible from the Open/Save toolbar buttons (how about the menu,
there, BTW. Does that show up, correctly?). I now tried moving that code
around, hoping a) it is the cause for this, and b) it is simply some
initialization order problem. If problem persists, you may want to test
disabling the two lines 
  proxy_import->setMenu...
  proxy_export->setMenu...
in rkward.cpp. If that helps, we can try to come up with a workaround.
 
> > > also in KF5:
> > >  - the valueslots in dialogs seem to be incapable of detecting the
> > >    class of selected objects, i only get "unknown" and a warning
> > > symbol
> > 
> > Which plugin? Works for me in "Basic Statistics" (match by class
> > name), and "t-test" (match by data type).
> 
> i see it in the IRT plugin (e.g., fit rasch model). does the code
> need fixes?

I see. But did that work in the KDE 4 version? Problem is you are
checking for a) data.frame or matrix class (that test succeeds), then
b) for data type "numeric". Problem is b), since RKWard considers
matrices and data.frames as mere "containers" that do not have a data
type. I can see your intention, and this should not be too hard to
implement. I am not sure, though, whether all existing plugins that do
test for numeric data, are prepared to handle matrices, and esp.
data.frames. Alternative would be to introduce a new notation for this
case, like "numeric_columns", "string_columns", etc.

Thoughts?

> > Note that I'll probably be disconnected from the net for most of
> > tomorrow (switching provider).
> 
> o-oh... hope you're not off for the next 6 months... 8-]

Keeping my fingers crossed...

Regards
Thomas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <http://mail.kde.org/pipermail/rkward-devel/attachments/20160914/448f64a0/attachment.sig>