Re: [Gpsbabel-code] [PATCH] Win32 installer for GPSBabel

[Robert L. <rob...@gm...>]

Hi, Ulf

On Nov 22, 2007 5:07 PM, Ulf Lamping <ulf...@we...> wrote:

> After some discussion on the openstreetmap ML, there seemed to be demand
> for a Win32 installer for GPSBabel (especially for the GUI part).
>

That's surprising to me.  We've averaed 5-600 downloads a day for a long
time
and I have only heard a very few of those requests.   About the time that
web browsers
and XP learned to handle zip files sensibly, the number of such requests
dropped to
a number statistically equal to zero to me.


First of all, is there a demand for such an installer (and there was
> only lack of time/knowledge just to do it), or do the gpsbabel
>

I'm not sure about the demand.   I remember investigating it long ago
and NSIS came up.  This was before the Microsoft installer became free(er?).

Can I cross develop NSIS packages?  I often build releases on a system other
than Windows.   So cross development matters a lot to me.  How well does
NSIS handle hosting on Linux or Mac?

- find a good place to put the NSIS sources in (put it under msvc/nsis,
> win32/nsis or simply nsis?)


As it's not bound to Visual C, I'd say the middle answer is the most correct
one.




> - better integrate the NSIS build into the source tree (e.g. I'll need
> to know where gpsbabel.exe is to be found after generation)


It depends on how the source tree was configured and built.



> - only .kpx file extensions are currently associated ("find" more file
> extensions that needs to be integrated into the windows explorer)
>

I don't know what KPX is.

I do feel that most users do not want to associate GPSBabel with all the
input files that it can handle, though, if that's what you're suggesting.

It would certainly violate the principle of least surprise for, say, a
mapsource
user to click on a gdb file and get GPSBabel instead of Mapsource.


> modem users), if you are interested in the installer.exe itself (~
> 800kB), please just drop me a note and I'll send it to you ...



Adding 800kB of overhead to a program that's currently only about 1000kB
is going to be a tough sell.

I'm not opposed to the idea, but I'm going to ask you to sell me on the idea
if you really think it's something we should adopt.

Thanx,
RJL