Thu 07 Nov 2013 07:13:23 PM UTC, comment #12:
I don't really think there is any good way to solve this problem. So I'm OK with closing it. We really just need a variable browser/editor.
|
Thu 07 Nov 2013 06:48:43 PM UTC, comment #11:
Oops, closed report when you wanted it open. I'm re-titling to reflect the open issue with ND-arrays, since alignment has been fixed.
|
Thu 07 Nov 2013 06:46:54 PM UTC, comment #10:
I think there is plenty of ways we could be clever and have a nicer display, but this will work well enough for the 3.8.0 release.
|
Thu 07 Nov 2013 06:24:08 PM UTC, comment #9:
OK, it would be good to list out the first few elements, but we don't have a syntax for that if the elements would span more than one "page". I.e., a 2x2x2 array is only 8 elements, so they could all be displayed, but there is no simple one-line representation for displaying them.
|
Thu 07 Nov 2013 06:23:09 PM UTC, comment #8:
Fixed the same precedence problem affecting the display of char matrices here (http://hg.savannah.gnu.org/hgweb/octave/rev/b951a8351fd7). Everything seems fine now, closing report.
|
Thu 07 Nov 2013 06:06:56 PM UTC, comment #7:
I just fixed the display of ND-arrays here (http://hg.savannah.gnu.org/hgweb/octave/rev/8fde5edabedd).
|
Thu 07 Nov 2013 05:59:27 PM UTC, comment #6:
Oh yeah, that horrible precedence problem. Sorry.
|
Thu 07 Nov 2013 05:56:18 PM UTC, comment #5:
I fixed the problem with the display of empty cells (http://hg.savannah.gnu.org/hgweb/octave/rev/28e9562d708b). Turned out too be quite easy, the stream operator '<<' was grabbing the results of the test from the tertiary operator '?' rather than the result "A" : "B".
|
Thu 07 Nov 2013 05:52:48 PM UTC, comment #4:
With your latest changeset things are a lot better. Multi-dimensional arrays, rand(2,3,4), still don't display anything in the Value column though.
Also, there has been a regression for empty cell arrays. They now display as '1', rather than '{}'.
|
Thu 07 Nov 2013 06:01:35 AM UTC, comment #3:
I checked in the following additional change:
http://hg.savannah.gnu.org/hgweb/octave/rev/1d2e709bbbda
This should improve the situation for printing scalar and matrix values with the format that is in effect at the command line. However, there are some differences. For example, try the column vecctor
At the command line, we scan all values and get one format for the whole matrix. But in the workspace viewer the format selection is done for each value separately. The reason for this is to allow us to print out the first few values of a matrix efficiently even if the matrix is large. If we had to scan the whole matrix for max and min just to get the format for a few values it could take too much time for the GUI. Maybe this could be improved in the future, but I think it is good enough for now. At least we get some values displayed. Rather than worrying too much about the formatting, it would be better to spend the time implementing a fully functional variable editor.
|
Thu 07 Nov 2013 03:04:24 AM UTC, comment #2:
Seems good on the alignment.
However, there has been a a few regressions. Complex variables now display differently. 1+2i use to display as "1 + 2i". Now it displays as "(1,2)".
ND-arrays such as rand (3,3,3) use to display as "...". Now there is nothing in the value column.
The value column used to respect the format that was in place at the command line. Now it is fixed at short. So,
used to switch the value field to "12.3000e+006"
|
Thu 07 Nov 2013 12:26:04 AM UTC, comment #1:
I checked in the following changeset:
http://hg.savannah.gnu.org/hgweb/octave/rev/49d573a1deda
|
Wed 06 Nov 2013 10:57:05 PM UTC, original submission:
The variables in the Workspace window are not aligned. It appears as if extra whitespace at the front of the "Value" string is not being trimmed.
Sample code:
The resulting Workspace view is attached as a .png file.
|