On 06/12/2010 12:20 AM, John McCabe-Dansted wrote:
On Sat, Jun 12, 2010 at 5:49 AM, Vincent van Ravesteijn<[email protected]> wrote:
It sound like a good idea. I don't think keytest is getting much
benefit from version control as it is. With commit access I could
submit 100 atomic changes without flooding lyx-devel with messages
about a tool that currently I think only I use.
Uhhm.. but then you're flooding the lyx-cvslog list.
To put things in perspective, that would work out to about 1 message a week.
As it is I think the last patch I sent was in January. At the moment
development is basically happening outside svn and I am releasing new
versions into SVN.
There are a few ways we could go:
1) I commit every atomic change. This could easily be limited to a
maximum of 1/week, and I don't work on keytest every week so it would
average a lot less.
2) Commit regularly; say once a month, if there is something to commit.
3) Commit as often as I do now.
4) Split keytest off from LyX.
You should commit when you have something worth committing. It's up to
you to judge when that is. I think this tool has been very useful.
I don't think we have discussed what the benefit of having keystest
in-tree is. LyX can run without keytest, and keytest can be applied to
software other than LyX, so it is not strongly coupled to LyX, and
splitting it off would be a possibility. I am thinking of using
keytest as a general tool to test any software that has a rich
keyboard accessible GUI, at which point it may make sense to split it
off from LyX. Keytest is at 150K, and AFAICT I am the only person who
uses it. At what point does the overhead of keeping keytest in the LyX
tree outweigh the benefit?
I don't see much actual overhead. Again, it's up to you if you want to
split it off.
Richard
