r.patch fails on Vista

[cmbarton]

Students in my class have just discovered that r.patch fails under WinGRASS. The GUI won't launch and it gives an error from the command line. The errors are none-specific in both cases, making it difficult to understand what is going wrong. This happened on RC4 and is still a problem on the new RC6 version.

Michael

[4everskiff]

Replying to cmbarton:

Students in my class have just discovered that r.patch fails under WinGRASS. The GUI won't launch and it gives an error from the command line. The errors are none-specific in both cases, making it difficult to understand what is going wrong. This happened on RC4 and is still a problem on the new RC6 version. Michael

Hi Michael,

could you please try again? because I just tried the command and it works fine. Thanks,

Marco

[cmbarton]

Marco,

It's failed on several student machines here. I don't have a windows box, but will ask others to try. One of the systems that it failed on is running Vista. I don't know if another one is Vista or XP.

Michael

[4everskiff]

Replying to cmbarton:

Marco, It's failed on several student machines here. I don't have a windows box, but will ask others to try. One of the systems that it failed on is running Vista. I don't know if another one is Vista or XP. Michael

Hi Michael,

I think that Vista is the key. I build and test WinGRASS on XP, so I cannot figure out how does it run on Vista. This links to an issue I already pointed out in the list: it's a fact that any professionist or scientist who works on Windows actually still uses xp or 2000, but the future is on Vista, since new machines (specially notebooks) seem to not work with xp, due to unreleased libraries for xp to manage new HD architectures. Should we start to think to build on both Vista and XP, and thus release two versions of WinGRASS? I think that, at least, we need to intensively test on Vista.

See also Glynn's point of view:  http://lists.osgeo.org/pipermail/grass-dev/2008-March/036820.html

Marco

[jrobins]

Marco,

I'm running winGRASS version 6.3.0RC6 on Vista. I configured GRASS with the windows installer program. I tried r.patch using the spearfish dataset, but when I try to launch r.patch from the wxPython GUI's menus, I first get a security alert which prompts me to allow or deny r.patch.exe access to my computer. When I press 'Allow', I get the following error message:

This application has failed to start because libgrass_gis.6.3.0RC6.dll was not found. Re-installing the application may fix this problem.

Then r.patch.exe crashes and Vista automatically kills it in the same manner that it would kill a hung program. The GUI seems to work fine afterwards.

When I run r.patch from the menu in the Tcl/Tk? GUI, I get this error:

couldn't execute r.patch: invalid argument.

Running r.patch from the Output-Gis.m window fails without an error message.

I hope this helps,

John

[cmbarton]

Yup. It's a Vista problem. I asked a student running XP to test it and it ran fine. Weird.

Michael

[mlennert]

I've never touched Vista, but maybe until we fix the code, people could try to use the XP compatibility mode [1], but I don't know how to tell Vista to apply this to all grass .exe files without having to click on every single one of them.

[1]  http://www.howtogeek.com/howto/windows-vista/using-windows-vista-compatibility-mode/

[jrobins]

I tried switching r.patch.exe to XP (SP2) compatibility mode, but it doesn't seem to help. I did, however, find libgrass_gis.6.3.0RC6.dll in my lib directory, so that isn't the problem.

[4everskiff]

Replying to jrobins:

I tried switching r.patch.exe to XP (SP2) compatibility mode, but it doesn't seem to help. I did, however, find libgrass_gis.6.3.0RC6.dll in my lib directory, so that isn't the problem.

Thanks John,

I also tried with wxPython GUI on XP, and r.patch works fine. As I figured out in past comments, this is only a Vista problem. I'm afraid. As I told some time ago, talking with Glynn, I think that we should start to think to build also on Vista (even if Glynn doesn't agree). This will be probably an available option on next month, when I'll buy a new PC (I guess on June).

Marco

Marco

[marcopx]

Replying to jrobins:

Marco, I'm running winGRASS version 6.3.0RC6 on Vista. I configured GRASS with the windows installer program. I tried r.patch using the spearfish dataset, but when I try to launch r.patch from the wxPython GUI's menus, I first get a security alert which prompts me to allow or deny r.patch.exe access to my computer. When I press 'Allow', I get the following error message: This application has failed to start because libgrass_gis.6.3.0RC6.dll was not found. Re-installing the application may fix this problem. Then r.patch.exe crashes and Vista automatically kills it in the same manner that it would kill a hung program. The GUI seems to work fine afterwards. When I run r.patch from the menu in the Tcl/Tk? GUI, I get this error: couldn't execute r.patch: invalid argument. Running r.patch from the Output-Gis.m window fails without an error message. I hope this helps, John

Hi John,

I just built GRASS 6.3.0 on Windows Vista (Home Premium) through the MSYS/MinGW environment, exactly as I built it on XP. I made only few tests, but both r.patch and v.patch worked fine for me.

This said, I don't think that it depends on the new build: I think that the problem lies on the UAC system, because it forbids third party apps to write outside the "Program Files" and "Users/Account? Name" folders, specially if the destination folder is directly in the unit root (such as C:\), and sometimes even if the user gives the permission after the UAC alert prompt message;

So, disable the UAC (see  http://www.vista4beginners.com/How-to-disable-UAC) and then try r.patch with the current WinGRASS binary release (still built on XP); Let me know if it works :-)

Marco

[hamish]

note dupe in #759. reopen this ticket (with all the discussion) if needed.

[cmbarton]

I searched the tracker for "r.patch" and didn't find anything. I guess it was because I wasn't searching closed tickets too. This definitely sounds like #759. I don't know what the protocol should be here.

Michael