[GRASS-dev] [GRASS GIS] #1428: WinGRASS + how to deliver Microsoft Visual C++ Redistributable Package (vcredist)?

[GRASS GIS]

#1428: WinGRASS + how to deliver Microsoft Visual C++  Redistributable Package
(vcredist)?
--------------------------+-------------------------------------------------
 Reporter:  dhastings     |       Owner:  grass-dev@…              
     Type:  defect        |      Status:  new                      
 Priority:  blocker       |   Milestone:  6.4.3                    
Component:  Installation  |     Version:  6.4.3 RCs                
 Keywords:  wingrass      |    Platform:  MSWindows 7              
      Cpu:  x86-64        |  
--------------------------+-------------------------------------------------

Comment(by hamish):

 Replying to [comment:45 hellik]:
 > Replying to [comment:42 hamish]:
 > >
 > > I think it just needs an extra [Yes,please][No,thanks] page at the end
 of
 > > the installer wizard with a download option like we already have for
 the
 > > sample datasets.
 >
 > are you sure? - regarding the mechanism behind the scene....

 I just mean the technique using wget/curl or whatever it is behind the
 scenes to download from a URL, uncompress, and install it. (with nice
 checks if there is no network connection or the download fails for some
 other reason)  Bonus would be md5sum checks on the downloaded file(s).

 It seems taken care of by NSISdl::download and untgz::extract in your
 patch, thanks.


 > done by r55817

 wording tweaked a little bit in r55819 to use the magic "system" library
 language.


 The idea to put the d/l in its own page at the end of the installer wizard
 instead of with the sample dataset downloads was because I worry it might
 get clicked past by someone hitting Next Next Next Next and not paying
 attention, then they wonder why it doesn't work.

 Another idea to avoid the registry key search- after figuring out which
 dlls are actually needed, we'd to compile a tiny program depending on them
 and have it run as part of the startup script and fail with a more
 informative error message (or, if we can't do much about the "this process
 ended" message, we can at least catch that it failed and the startup
 script continuing on from there sees the process return code and aborts
 with a more informative error message.) ?

 Our next step seems to be some quality time spent exploring with
 Dependency Walker..


 thanks for the patch!
 Hamish

-- 
Ticket URL: <https://trac.osgeo.org/grass/ticket/1428#comment:48>
GRASS GIS <http://grass.osgeo.org>