Wed 24 Jun 2015 09:08:14 PM UTC, comment #17:
I committed the original change which eliminates the use of the "reverse" keyword and just flips the axis limits (http://hg.savannah.gnu.org/hgweb/octave/rev/977853866d69). This is the patch from Dan in comment #10. Closing report.
|
Tue 23 Jun 2015 04:39:11 PM UTC, comment #16:
I think the dashed line for gnuplot 5.0 should be a separate bug report. There is also a related changeset for the "surf" function in which I made the line color/thickness/dash operable in 5.0+ using a has-feature "dashtype":
https://savannah.gnu.org/bugs/?34523
Perhaps all the uses of has-feature "term option dashed" could be replaced by has-feature "dashtype".
|
Tue 23 Jun 2015 10:26:58 AM UTC, comment #15:
Sorry here is not proper place to write new dash/dotted feature of gnuplot 5.0. I will transfer the previous writing somewhere else in maintainers ML.
|
Tue 23 Jun 2015 09:53:27 AM UTC, comment #14:
NOTES for ver. 5
The dot/dash pattern of a line can now be controlled independently from other line properties using the keyword "dashtype".
terminal option dasded is deprecate on Ver. 5.
I have a quick look at _gnuplot_drawnow_.m of ver. 4.0.0.
has_termoption_dashed does not have mean on gnuplot 5.0.
If one want support dash and dotted type on gnuplot ver. 5, the routine should be changed entirely.
|
Tue 23 Jun 2015 07:35:00 AM UTC, comment #13:
From gnuplot 5.0 NOTES,
The "reverse" keyword (e.g. "set xrange [:] reverse") now affects only autoscaling. It has no effect on explicit ranges. "set xrange [0:1] reverse" is not the same as "set xrange [1:0]".
|
Tue 23 Jun 2015 07:10:06 AM UTC, comment #12:
gnuplot 4.7
For gnuplot,
if the minor version number in odd (1, 3, 5, 7, 9), the version is development branch. The gnuplot 4.7 exist before ver 5. release but it deprecated.
The last version gnuplot 4 is the 4.6.7.
http://sourceforge.net/projects/gnuplot/files/gnuplot/4.6.7/
It is a bug fix release and only source release.
There is no plan to update release in gnuplot 4.
|
Sun 24 May 2015 05:04:31 AM UTC, comment #11:
Doesn't seem to cause any problems with gnuplot 4.6.6. Should this go on stable or default?
|
Sun 24 May 2015 04:55:14 AM UTC, comment #10:
Updated changeset attached...
(file #34085)
|
Sun 24 May 2015 12:48:13 AM UTC, comment #9:
@Dan: I'm running gnuplot 4.6.6 so I can test things. The original cset no longer applies cleanly to the source tree. Can you update it?
|
Sat 23 May 2015 12:27:03 PM UTC, comment #8:
Unmarking as postponed and retitling since gnuplot 5.0 has been out for a while now, and there may never be a 4.7.
|
Sat 23 May 2015 06:36:06 AM UTC, comment #7:
Forgot about this one from long ago...
I don't think __gnuplot_has_feature__ is necessary in this case as < 4.7 appears to accept the limits being swapped and no "reverse/noreverse" specifier. I'm guessing the "reverse/noreverse" was removed from gnuplot because of the confusion that can result from syntax like this:
Should the "reverse" change the range to [0:4]?
If someone else wants to confirm that 4.6 works with the original change, that would save having to detect the feature. (Which is a slight problem, that I'm going to post to gnuplot discussion, but isn't important here.)
|
Mon 28 Oct 2013 03:05:57 AM UTC, comment #6:
Marking as postponed since gnuplot 4.7 is not yet released.
|
Thu 20 Jun 2013 06:18:26 PM UTC, comment #5:
Thanks. I'll follow up when 4.7 is released, but if I miss it feel free to apply your change set.
|
Thu 20 Jun 2013 12:08:09 AM UTC, comment #4:
Dan, I modified your changeset slightly to use the function _gnuplot_has_feature_ which is our standard for querying whether gnuplot supports something. I'm attaching it to this bug report and you can use it if gnuplot does in fact go forward with their proposed modifications in 4.7.
(file #28360)
|
Wed 19 Jun 2013 10:56:39 PM UTC, comment #3:
Yes, it is particular to version 4.7 which has or hasn't come out yet. (I'm using the gnuplot development version on my system.) The behavior of "reverse" is going to change because it had a property of stickiness meaning once it was set, it wasn't possible to unset it. I eventually recalled having wondered about this on the gnuplot discussion list:
http://marc.info/?l=gnuplot-info-beta&m=134712826931471&w=2
http://marc.info/?t=124817327100003&r=1&w=2
I would have pursued this particular bug report (bug #38914) if I had felt comfortable with the change gnuplot developers are making, but the proposed change just didn't seem satisfactory so I'm just sort of taking a wait-and-see approach whether this is the final change when 4.7 comes out. 4.7 hasn't been released yet.
The attached patch does check for the version:
+ use_reverse_keyword = compare_versions (_gnuplot_version_ (), "4.7", "<");
so it is ready to go if gnuplot 4.7 is released and has the currently proposed change to "{no}reverse".
|
Wed 19 Jun 2013 10:18:36 PM UTC, comment #2:
Is this particular to a specific version of gnuplot? I am using gnuplot 4.6.2 and the axis is correctly reversed for me with the existing code.
My test was
|
Tue 07 May 2013 12:45:30 AM UTC, comment #1:
As I was preparing the changeset I found a similar patch already on my computer. I must have fixed this a half year ago but didn't get around to submitting the changeset because there was a lot of other activity at the time with getting the GUI more in line with the signal/slot concept. I can't find any bug report that I created around the time of the original patch.
(file #28037)
|
Tue 07 May 2013 12:20:05 AM UTC, original submission:
The "image" command is not producing an image with the y-axis reversed, although the matryoshka pointer info all looks correct.
The issue is that gnuplot graphics code is using the keyword "reverse"/"noreverse" but that only applies to automatic xrange specifications. When actual numerical values are given for the range (which is always the case here), the range values (e.g., xlim) need to be reversed.
I will attach a patch after submitting this report.
|