[Kst] Sort of serious bug...

[Barth Netterfield]

Ben has discovered a pretty serious bug (thanks!):

commit a319703a76e27b30b3b6c122c708583132421939 added scalars 
iMax and iMin to vectors.

This changed the number of scalars that a vector has.

This, in turn, breaks scalar lookup from all kst files created before that 
commit.

*Which is super bad.*

The solution is not trivial, and reveals a fundamental fragility in the 
shortName system combined with how we save and load objects.

Options: 
1) revert the commit and declare that no new scalars can ever be added 
to vectors

2) declare that all files from before the commit are now incompatible with 
kst-current

3) disable iMax and iMin for vectors in files saved before the commit (even 
if they are re-saved from a new version).

4) Come up with a more elaborate hack work around in the file reader or in 
vector-scalar naming to bandage over the fundamental fragility.

Not a big fan of (1).
I think we can agree that (2) is unacceptable.
(3) might be easy, but produces surprising behavior (namely, that these 
new features requires rebuilding the kst session)
(4) seems decidedly non trival as well as seeming hacky.

hmmm....
Sigh...
-- 
Barth Netterfield
University of Toronto
416-845-0946
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kst/attachments/20150304/38a36f90/attachment.html>