[GRASS-user] g.manual tries to run C:\Program Files\Internet Exporer\iexplore (the 'L' is missing from 'Explorer')

[Hamish]

Glynn wrote:
> Can you modify the script to print its arguments? And
> g.manual to print out $GRASS_HTML_BROWSER.

note1 that until Maris's space-in-filename fixes in 6.5svn have been
reviewed & backported to 6.4 we may just be chasing things already
fixed and not being efficient in our use of time.

note2 until osgeo4w has a new release, diito with anything having
to do with that version.

note3 (a contingent of) the MSys devs has made it clear that they
won't even apply patches fixing spaces in pathname issues because
if it works at all they worry about being flooded with bug reports
for every GNU util they ship with etc which fails with that. I
don't really agree with that strategy, but it's their project so
whatever. I've got a conglomerate patch in their SF bug tracker
solving some core spaces in pathnames problems which Colin applies
(plus his own improvements). But AFAIK that's just the startup
stuff.

For GRASS_HTML_BROWSER, it is read in a number of contexts by e.g.
by g.manual shell script (6.x), g.manual python (7.x), wxGUI,
tcl/tk GUI, grass-run.bat, gui startups, ...

so it needs to be both cross-platform, cross-language, and cross-
environment (msys vs dos). thus IMO the only portable solution will
be to store it in the OS's native FS format and have each client
code translate downstream as needed rather than guess that up
front.

"C:\Program Files" is not portable due to i18n, 
the above-mentioned MSys patch has long -> 8.3 DOS filena~1
batch file translation magic for reference, I assume python has
all the tools you need built in, tcltk can be pursuaded, ...
which leaves msys's /bin/sh.

I find msys to be too large complex to simply say "msys does or
does not support this or that" as a blanket statement.


Hamish