After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 530711 - Change in array eval between 1.6 and 1.8
Change in array eval between 1.6 and 1.8
Status: RESOLVED WONTFIX
Product: Gnumeric
Classification: Applications
Component: Analytics
1.8.x
Other All
: Normal normal
: ---
Assigned To: Jody Goldberg
Jody Goldberg
Depends on:
Blocks:
 
 
Reported: 2008-04-30 10:32 UTC by Pat Gunn
Modified: 2009-05-16 02:56 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
Spreadsheet which works in older versions of gnumeric (1.66 KB, application/octet-stream)
2008-04-30 10:34 UTC, Pat Gunn
Details

Description Pat Gunn 2008-04-30 10:32:48 UTC
Gnumeric 1.6.x allowed summaries
over computed values, e.g. SUM(A1:A9/B1:B9) or SUM(IF(B1:B9,A1:A9/B1:B9)). Gnumeric 1.8.x does not. This is a fairly serious (and undocumented) feature regression.
Comment 1 Pat Gunn 2008-04-30 10:34:08 UTC
Created attachment 110153 [details]
Spreadsheet which works in older versions of gnumeric
Comment 2 Pat Gunn 2008-04-30 10:36:36 UTC
Previously reported on Fedora's Bugzilla as https://bugzilla.redhat.com/show_bug.cgi?id=444566

Fedora9 preview includes Gnumeric 1.8.2, previous versions of Fedora used Gnumeric 1.6 or below, and I am requesting that they include Gnumeric 1.6.x instead of 1.8.x with the full release because of this problem.
Comment 3 Jody Goldberg 2008-04-30 11:23:00 UTC
Have you tried entering the values as array expressions (using ctrl-shift-enter rather than enter to complete an edit).

=sum(a1:a5/a1:a5) == 5 when entered as an array.
Comment 4 Pat Gunn 2008-04-30 11:55:14 UTC
This works admirably for sum(), but min() and max() both return wrong values (it seems to be using min as a selector for the first value and max as a selector for the second, rather than actually doing either operation). I suspect this is a bug in the array expression parser.
Comment 5 Pat Gunn 2008-04-30 12:01:19 UTC
nevermind, things seem to work ok with array evaluation now
Comment 6 Morten Welinder 2008-04-30 15:36:25 UTC
We did some array evaluation fixes sometime between 1.6 and 1.8.
Comment 7 Jody Goldberg 2008-04-30 16:11:01 UTC
The XL compatibility fixes appear to have broken existing sheets.  We need to look for a way to support the older semantics.  Possibly an internal eval flag and some xml extensions.
Comment 8 Andreas J. Guelzow 2008-04-30 20:51:15 UTC
Jody: those fixes had also broken some of the analysis tools that used the old semantics.
Comment 9 Pat Gunn 2009-02-26 16:42:40 UTC
Any changes on this?
Comment 10 Morten Welinder 2009-02-26 18:24:49 UTC
Realistically it will not happen.
Comment 11 Andreas J. Guelzow 2009-02-26 20:39:54 UTC
Morten: so should this bug be closed?
Comment 12 Morten Welinder 2009-05-16 02:56:27 UTC
--> WONTFIX.

We will, however, help with fixing up sheets that need it.
Just drop by #gnumeric.