[GRASS-dev] Re: [GRASS-SVN] r49533 - grass/branches/develbranch_6/gui/scripts

[Hamish]

Sorry to drag this on,

Martin:
> I simply don't understand why to waste energy with that in GRASS6.

- to preserve design consistency with earlier grass6 addon use.
- g.extension is our newest core infrastructure module, it is just now
  becoming mature, and what we do with this now will be in use for
  a long time to come. (for however long g6 is used)
- it itches


> to understand - did you request RC3 for the reason to make
> `g.extension` "to be tidy citizen" in G642 or for other reason?
> In the case of "tidy citizen your changes were introduced only in
> `devbr6`.

I would like to backport these changes to 6.4 in time for 6.4.2.
To ensure that all is well with that a quick rc3 would be nice
once it is done. I have not backported it out of the devel branch
yet as I hoped for some testing feedback and better consensus in
this discussion -- I don't like using the stable branch for
testing, and I don't like getting into the game where the last
person to sneak in their commit before release gets their way.

Perhaps that makes this drag on way too long and eat our energy,
but I think it is still better than the alternative way of shooting
first and asking questions later.

If there is nothing more I'll do that commit tomorrow.


> It's just complicates the situation before releasing 6.4.2.

better a little pain for us few now than to change it after
widespread release and cause pain for many users later on.


> The fact is that you call `make install` in `g.extension` and then
> you move the files around with one exception (html file).

I was mistaken when I earlier suggested that 'make install' was only
used for system-wide installs. That was used in an earlier iteration
or some uncommitted experimentation in the script, but it is not
what's in svn now.  None the less, either way and whatever percentage
it is, it gives the same result of moving 3 files then removing some
directories if they are empty ...


> I am very surprised that you are able to spend with hacking
> `g.extension` in G6 so much time.

shrug, it's my time and this is something I'd like to see us do really
well. The new wxGUI extension tool is really nice and useful, I would
like to ensure that the backend install is the same. I don't like to
make you spend your time on things you'd rather not, so I again offer
any help you need to make this work.


> You have decided how the things will be at the beginning.

Please don't think that I am so stubborn that I can never have my
mind changed in the face of solid reasoning. I am much more
concerned with finding the correct answer than personally being
always correct or getting my own way. Believe me, I would not give
you such a robust argument if I didn't respect your views or
talent and ability to show me what I hadn't considered. If after
that my ideas still haven't fallen, then I have more confidence
that they are not so silly.


> 2) important thing is RC3 - you have requested that, but it's not
> clear (at least for me) why and what you would like to do with
> `g.extension` in G64.

My plan is to backport the g.ext .sh + .py scripts that are now in
6.5svn, then if there is the will to do so put out a rc3, which if
everything works in we can release as 6.4.2 final soon after. If we
got plenty of testing in 6.4svn I guess a rc3 wouldn't be needed,
but I suspect we'd have the same people testing as who would for
6.5svn, so a rc3 might get a wider audience. shrug

(Also I would s/GRASS_ADDON_PATH/GRASS_ADDON_BASE/ in trunk)



best,
Hamish