Fri 24 Mar 2017 05:22:06 PM UTC, comment #14:
Changing status to fixed.
|
Mon 13 Mar 2017 12:07:30 AM UTC, comment #13:
Great. I pushed your patch here http://hg.savannah.gnu.org/hgweb/octave/rev/a921b6b772f3.
In the process I found a bug with clf ("reset") and FLTK which I will file separately.
|
Sun 12 Mar 2017 08:51:57 PM UTC, comment #12:
I attached a patch.
(file #39983)
|
Sun 12 Mar 2017 06:22:25 AM UTC, comment #11:
Actually, it can be any toolkit.
|
Sun 12 Mar 2017 06:20:24 AM UTC, comment #10:
I'm re-opening the bug report. With the gnuplot toolkit I get warnings whenever reset is used explicitly.
Sample Code:
Here is another example.
|
Thu 23 Feb 2017 06:44:40 PM UTC, comment #9:
I finally opted for disabling warnings directly in F__get__:
http://hg.savannah.gnu.org/hgweb/octave/rev/a88575e87481
|
Wed 22 Feb 2017 05:56:04 PM UTC, comment #8:
I just noticed that when using gnuplot graphics toolkit we get a warning each time an axes is redrawn:
This is because _gnuplot_draw_axes_ uses the special "__get__" function to obtain hidden properties, which triggers the warning.
It seems to be in one place only so silencing the warning there should be enough. I'll prepare a cset ASAP
|
Mon 20 Feb 2017 05:30:50 AM UTC, comment #7:
I pushed the patch here http://hg.savannah.gnu.org/hgweb/octave/rev/5d9d756724cc. Re-closing report.
|
Sun 19 Feb 2017 08:23:59 PM UTC, comment #6:
Confirmed. I attached a patch in which I disabled the warning as suggested by Markus.
(file #39793)
|
Sun 19 Feb 2017 06:53:43 PM UTC, comment #5:
"test imshow" now fails. The actual issue originates in reset where the deprecated properties raise a warning when they are set to their default value (graphics.cc @ line 5001).
Maybe that could be solved by turning that particular warning off at the beginning of the reset function and re-setting it to its previous setting afterwards (@ around line 10228).
Re-opening for this issue.
|
Sun 19 Feb 2017 01:04:05 AM UTC, comment #4:
I pushed your patch here (http://hg.savannah.gnu.org/hgweb/octave/rev/239a9fed80de). Only two small changes, in the .tst file I used '##' rather than '%%' to start comment lines and I found there were some lines that were missing semicolons.
Thanks. Marking as fixed and closing report.
|
Fri 17 Feb 2017 12:15:53 PM UTC, comment #3:
May I push the patch?
|
Thu 16 Feb 2017 05:16:07 PM UTC, comment #2:
>> How does Matlab handle this?
I don't really know but I am under the impression (see comments in bug #50227) that they first hide the property, eventually announce the change in the release notes (not sure), and remove it later if needed.
I think the addition of a warning is much cle
|
Thu 16 Feb 2017 04:21:43 PM UTC, comment #1:
Deprecation is definitely a nice way to handle obsolescence. How does Matlab handle this? Do they just remove properties, and then people fix their scripts when Matlab produces an error?
|
Thu 16 Feb 2017 01:13:49 PM UTC, original submission:
We currently don't have a way to deprecate properties as we do for built-in functions.
The attached patch is a first attempt to do so:
- add a way to mark a property as deprecated (new "d" flag for genprop.awk) so that users are automatically warned when they us it
- add a .tst file in which we can check that a given property has disappeared from a given version.
|