Re: Item groups in bug tracker: "incorrect result" too broad, and masked by "regression"

[LachlanA]

Greetings Michael,

I'm certainly not saying that bugs in plotting are necessary not serious
bugs.  All was saying is that I would like this category to exclude them,
and even for that, you have almost convinced me otherwise.

If a plot looks "right", but plots the wrong numbers, that possibly belongs
in the same category.  I mainly wanted to exclude bugs like the top 8 open
bugs from   Category=Plotting  and  Item Group = Incorrect Result:

#46590 wrong result using pbaspect ()
#46469 Changing papersize and paper position leads to cropped image
#46325 Error with rlocus plot  [plot disappears after autoscale)
#46263 fill plot freezes and produces no result
#45991 Command Window crashes after trying to plot
#45850 Minor grid lines in wrong locations when using manually set xtics
#45631 Line dissapear from plot when modifying "plotboxaspectratio"
#45469 Program execution stops unpredictably with error in plot statement

Although some of these are serious, like the command window crashing, it is
of a different nature because it is detectable.

Number 9 may be relevant:
#45440  Lines are wrong during a Z+ zoom and after.
although if it doesn't affect the exported eps/pdf/whatever then I'd still
be inclined to exclude it.

Then we have another bunch that I'd exclude from "Wrong answer", namely

#44776 [Qt] Restore window button does not trigger a repaint event for its
content
#44410 "axes (hax)" should reorder axes/colorbars/legends among uipanel
children
#44409 Reparenting the current xes under an ui panel makes it not current
#44323 Zooming in 3E plots works on XY plane rther than projection plane
#43651 special character (diameter sign) in plot produces warning or doesn't
show up
#42940 Changing all fontsizes in plot gives warning when legend present
#42884 Property FontName not changeable before image export

Then "#42766 incorrect color patches for countourf with NaNs in data" is
probably a "wrong answer" (or whatever), but I'd possibly exclude it on the
basis of being an NaN corner case.  (Again, this is just my opinion, which
I'm putting out to start a discussion.)

Browsing "Category=Octave Function" and "Item Group=Incorrect Result" also
gives results that are mostly not of the category that I was proposing. 
(Perhaps they are actually rare bugs, or perhaps they just get fixed quickly
with the current categories.  Either way, it may be that there is no need
for a new or renamed category.)  Some open bugs that are definitely in scope
are:
bug #43702: interpn(X1,X2,F,iX1,iX2"spline") yields incorrect results when
iX1,iX2 are matrices
bug #41454: Inaccurate/incorrect result for inverse of symmetric matrix
bug #39014: Wrong determinant for some (large) matrices

Closed bugs include
bug #45739: sort 'descend' together with vectors created with the colon
operator
bug #45455: quantile returns sorted vector when called on singleton
dimension
bug #44071: ismember fails with unsorted complex set
bug #43369: rats has some internal overflow  [even though it is an incorrect
string rather than incorrect number]
bug #42485: issymmetric always false for logical matrix
bug #42418: luupdate transposes permutation matrix if not constructed
properly

Cheers,
Lachlan



--
View this message in context: 
http://octave.1599824.n4.nabble.com/Item-groups-in-bug-tracker-incorrect-result-too-broad-and-masked-by-regression-tp4674355p4674386.html
Sent from the Octave - Maintainers mailing list archive at Nabble.com.