First ,
I must apologize for my intermittent communication, I am not able to check email regularly this week.
I downloaded the most recent standalone installers tonight and the bug in 7.8.6 https://github.com/OSGeo/grass/issues/1965 is still there in the WinGRASS-8.1.dev-7e9610cfd-28-Setup.exe and WinGRASS-8.0.dev-ffb060a9c-36-Setup.exe binaries.
I see [https://github.com/OSGeo/grass/pull/2068](https://github.com/OSGeo/grass/pull/2068) has been merged into 8.0 release, but for some reason,
this does not seem to be fixing these standalone builds. Since this is not an issue with the core grass code,
but with the mswindows/GRASS-Installer.nsi.tmpl file, is there the possibility that the .nsi.tmpl file is not
being refreshed during the build process?
This all boils down to permissions on the etc/grass80.py file only set to read/execute for System and Administrative groups,
not for local users. All of the other python files in the etc directory of the windows installation have the correct read/execute permissions ,
as set in line 594 of the mswindows/GRASS-Installer.nsi.tmpl , which was not changed in pull request 2068.
Pull request 2068 simply changed line 762 to match line 594 to give the same permissions that all of the other python files in
the etc directory. This worked on the version compiled on my development computer from source. Unfortunately,
I will not be able to compile the current source code until I get back to my development computer late this week.
Grass 8.0 RC2 windows standalone installer has the same issue.
Doug
On Mon, Jan 17, 2022 at 12:52 AM Doug Newcomb <gistinker@gmail.com> wrote:
First ,
I must apologize for my intermittent communication, I am not able to check email regularly this week.
I downloaded the most recent standalone installers tonight and the bug in 7.8.6 https://github.com/OSGeo/grass/issues/1965 is still there in the WinGRASS-8.1.dev-7e9610cfd-28-Setup.exe and WinGRASS-8.0.dev-ffb060a9c-36-Setup.exe binaries.
I see [https://github.com/OSGeo/grass/pull/2068](https://github.com/OSGeo/grass/pull/2068) has been merged into 8.0 release, but for some reason,
this does not seem to be fixing these standalone builds. Since this is not an issue with the core grass code,
but with the mswindows/GRASS-Installer.nsi.tmpl file, is there the possibility that the .nsi.tmpl file is not
being refreshed during the build process?
This all boils down to permissions on the etc/grass80.py file only set to read/execute for System and Administrative groups,
not for local users. All of the other python files in the etc directory of the windows installation have the correct read/execute permissions ,
as set in line 594 of the mswindows/GRASS-Installer.nsi.tmpl , which was not changed in pull request 2068.
Pull request 2068 simply changed line 762 to match line 594 to give the same permissions that all of the other python files in
the etc directory. This worked on the version compiled on my development computer from source. Unfortunately,
I will not be able to compile the current source code until I get back to my development computer late this week.
Hi,
po 17. 1. 2022 v 6:53 odesÃlatel Doug Newcomb <gistinker@gmail.com> napsal:
I must apologize for my intermittent communication, I am not able to check email regularly this week.
I downloaded the most recent standalone installers tonight and the bug in 7.8.6 https://github.com/OSGeo/grass/issues/1965 is still there in the WinGRASS-8.1.dev-7e9610cfd-28-Setup.exe and WinGRASS-8.0.dev-ffb060a9c-36-Setup.exe binaries.
I checked the latest 8.1 build
https://wingrass.fsv.cvut.cz/grass81/WinGRASS-8.1.dev-d21dd0a03-35-Setup.exe,
the relevant commit is included [1]. The attached file has been used
by packager to create an exe installer.
Martin
[1] https://github.com/OSGeo/grass/commit/7fca178f18a68ed7d61367d0084733163015eb82
--
Martin Landa
http://geo.fsv.cvut.cz/gwiki/Landa
http://gismentors.cz/mentors/landa
(attachments)
GRASS-Installer.nsi (36.2 KB)