Bug report #9203
R Plots are blank on qgis64bit
Status: | Closed | ||
---|---|---|---|
Priority: | High | ||
Assignee: | Victor Olaya | ||
Category: | Processing/Core | ||
Affected QGIS version: | 2.4.0 | Regression?: | No |
Operating System: | Easy fix?: | No | |
Pull Request or Patch supplied: | No | Resolution: | |
Crashes QGIS or corrupts data: | No | Copied to github as #: | 17831 |
Description
The 'results viewer' shows empty 'R plots'. This problem is very similar to what was reported 1 year ago (Bug #5882), where we can get shapefiles and rasters as outputs (eg Minimum Convex Polygon - 'Home Range Analysis' group), but not plots generated by R. A good example is when we tried the 'G Function' in the 'Point Pattern Analysis' group that is available as a default R script.
In the temporary folder (or set output folder), we get two files: *.html and *.html.png. The *.html.png produces a very small image (~5kb) of the plot. This plot cannot be seen or copied from the 'results viewer'.
Details:
Windows 8.1 (x64)
QGIS 2.0.1 (x64)
Processing 2.0-20131120
R 3.0.2 (x64)
Related issues
History
#1 Updated by Giovanni Manghi almost 11 years ago
- Status changed from Open to Feedback
duplicate of #7643 ?
#2 Updated by Raphael Vanderstichel almost 11 years ago
- Assignee changed from Victor Olaya to Giovanni Manghi
The issue does seem to be similar to #7643, however, there are no solutions yet. From comments in #7643, I have tried installing an older version of R (2.15.3) and re-assigning the R folder in the 'processing' options to R 2.15.3, but this did not solve the problem. I also looked into the \\.qgis2\\processing folder to find potential 'rlib' files, but these are no longer here and I suspect this changed after the move from SEXTANTE to 'processing'. Another suggestion was to use the package R2HTML - this I am not so keen on because it will mean modifying every R script that requires a graphs.
I am willing to try other suggestions.
#3 Updated by Filipe Dias almost 11 years ago
I think this bug is specific to the 64 bits version. Can you retry with the 32 bits version?
#4 Updated by Raphael Vanderstichel almost 11 years ago
Yes, I can confirm that this problem is specific to the 64-bit version of QGIS. I ran both versions (32 and 64 bit) in a virtual machine (Win 7 x64) and had no issues with the 32-bit version, but repeated the issue in the 64-bit version. I also tried the 32-bit version of QGIS in a virtual machine with Win 8.0 x64 and had no problems plotting graphs, while it was with my 64-bit version of QGIS in Win 8.1 x64 that I first noticed the issue.
Unfortunately, I cannot switch to the 32-bit QGIS (Win 8.1 x64) because I ran into a more serious issue with GRASS, where it would simply crash after the first operation (in GRASS (standalone), within QGIS, and within the Processing plugin).
I am still willing to try solutions within virtual machines to help with the debugging of the plotting issue within the Processing plugin. I really like this plugin and think this is a necessary tool for spatial analysis.
#5 Updated by Giovanni Manghi almost 11 years ago
- Assignee changed from Giovanni Manghi to Victor Olaya
- Subject changed from R Plots are blank to R Plots are blank on qgis64bit
- Status changed from Feedback to Open
Filipe Dias wrote:
I think this bug is specific to the 64 bits version. Can you retry with the 32 bits version?
does this happen only on Windows or Linux too?
#6 Updated by Giovanni Manghi almost 11 years ago
Unfortunately, I cannot switch to the 32-bit QGIS (Win 8.1 x64) because I ran into a more serious issue with GRASS, where it would simply crash after the first operation (in GRASS (standalone), within QGIS, and within the Processing plugin).
this is another issue. If it GRASS standalone crashes for some reason then it is normal it does not work in the qgis plugin or processing toolbox.
How GRASS was installed, using the qgis standalone installer? osgeo4w? what is the module that makes it crash?
I really like this plugin and think this is a necessary tool for spatial analysis.
yes, in fact it is not a plugin anymore, is part of qgis now.
#7 Updated by Filipe Dias almost 11 years ago
does this happen only on Windows or Linux too?
On Linux it works as expected.
#8 Updated by Raphael Vanderstichel almost 11 years ago
I haven't tried it on Linux, nor will I in the near future.
I had installed QGIS as a standalone (not osgeo4w). Every command made it crash. Strangely though, I could load layers from the database, but as soon as I ran a command, a windows shell window opened and closed quickly (could not read any text within it), and then it stopped responding. Examples of commands I tried - v.buffer.dist and v.distance.
#9 Updated by Giovanni Manghi almost 11 years ago
I had installed QGIS as a standalone (not osgeo4w). Every command made it crash. Strangely though, I could load layers from the database, but as soon as I ran a command, a windows shell window opened and closed quickly (could not read any text within it), and then it stopped responding. Examples of commands I tried - v.buffer.dist and v.distance.
processing works overall ok on Windows standalone/osgeo4w, and what you are experiencing can be a local issue. A screencast would help try understand what it is going on.
#10 Updated by Filipe Dias over 10 years ago
- Priority changed from Normal to Severe/Regression
Since this prevents users to use some R scripts in QGIS 64 bits I'm changing the priority to Blocker.
#11 Updated by Giovanni Manghi over 10 years ago
- Priority changed from Severe/Regression to High
Filipe Dias wrote:
Since this prevents users to use some R scripts in QGIS 64 bits I'm changing the priority to Blocker.
I'm changing this to non-blocker for different reasons:
a blocker means a regression, this should not be the case, and then processing can be always updated as plugin update, so I don't think that even a huge issue should block a qgis release.
#12 Updated by Filipe Dias over 10 years ago
I believe this qualifies as blocker because it is a functionality that doesn't work as expected. It is not technically a regression because it never worked properly, but it's a problem that usually frustrates users and leads them to conclude "the R integration doesn't work".
#13 Updated by Giovanni Manghi over 10 years ago
Filipe Dias wrote:
I believe this qualifies as blocker because it is a functionality that doesn't work as expected. It is not technically a regression because it never worked properly, but it's a problem that usually frustrates users and leads them to conclude "the R integration doesn't work".
I understand Filipe, but it is not a regression and when fixed it can always be updated as a plugin. In my point of view then processing issues should never be blockers, especially if during the pre-release period there is now one actively working on them.
#14 Updated by Giovanni Manghi about 10 years ago
- Project changed from 78 to QGIS Application
- Crashes QGIS or corrupts data set to No
- Category deleted (
66) - Affected QGIS version set to 2.4.0
#15 Updated by Giovanni Manghi about 10 years ago
- Category set to 124
#16 Updated by Håvard Tveite over 9 years ago
The fix reported in #7643 works for me with QGIS 2.8.1 from OSGeo4W64.
After deleting the processing directory in .qgis2/python/plugins,
plots were generated as expected by for instance the "F Function" and
the "G Function".
#17 Updated by Filipe Dias over 9 years ago
- Status changed from Open to Closed
#18 Updated by Giovanni Manghi over 7 years ago
- Regression? set to No
#19 Updated by Giovanni Manghi over 7 years ago
- Easy fix? set to No
#20 Updated by Giovanni Manghi over 7 years ago
Some providers are being removed from QGIS/Processing (will be available as plugin) and so are their categories in the bug tracker. To not leave them orphaned of a category they are being reassigned to processing/core.
#21 Updated by Giovanni Manghi over 7 years ago
- Category changed from 124 to Processing/Core