Re: [glob2-devel] glob2-0.8.22: What went wrong. How to improve. What to do now.

[Kai Antweiler]

>>  - Put up a new tarball in the wiki and a note that this one is
>>    more portable.
>>    (I created a tarball, but I don't know how to upload.)
>
> Do you have ssh access on ww.globulation2.org ?

I don't know.  So probably not.


> I agree with the rest of your points. I think we should concentrate on fixing 
> the bugs discovored lately and produce a new release soon enough before our 
> reputation goes down too much.

I think this would be best from a software company point of view.
But it won't be realistic for us.

Here is our cvs head line changes statistic for this year so far:

genixpro              23724 ************************************************
nct                    2969 ******
giszmo                 1032 **
asayed                  963 *
donkyhotay              276 
snth                     93 
convert-repo             27 
retardedpope              6 
appleboy                  4 

convert-repo: hasn't commited anything.  It is a script that converts
              git repositiories to mercurial.
retardedpope: transation only, so don't expect bug fixes
asayed:       Haven't heard from him since february.
snth:         busy
donkyhotay:   busy


Bradley will have a new map header implemented before the bugs get fixed.

But there is a bug that is in my opinion really nasty and must be fixed
as soon as possible.  And when it's fixed, it's time to do a release.
https://savannah.nongnu.org/bugs/?19542

I think that we can fix this bug in alpha22 once we have found it in
HEAD.  And then make a bug fix release based on alpha22.
I therefore think using HEAD exclusively for now is best,
if we consider how our development realistically will look like in the
next months.

But we must upload a tarball that allows our AMD users and the users of
distributions that don't store *.m4 files in the same directory as it
is done in OS-X to install glob2, now.
Or alternatively point them to our alpha21 release.
In our alpha22-rc cvs repository this is fixed.


> In general, we should take more care at reported bugs and really try to 
> eliminate them.

I tried too, but there weren't many bugs in the week before our
release reported.  And no new serious bugs at all.


> On the feature side, I plan to allow myself to improve the look of
> the gui as this is well localised code that hadn't had any serious
> bugs for years.
>
> What do you think ?

I think that your gui changes are the most important improvements in
glob2 as long as I know it (i.e: 2004-2007).
They simply make a difference.  Users don't expect open source games
to have a good looking gui.  And it's the first thing that you
recognize when you try a game.  If the gui doesn't look well, or
behave intuitively most users won't give the game a chance.
Also the gui is visible on screenshots on our pages and some game reviews.

It's a pity that the qt rewrite is stalled.  I would be a big plus
considering the intuition.  Users would get an interface that they
are already used to.

So direct answer to your question:
Continue your work.  But we should keep alpha22-rc untouched as long
as we don't have a non-sticky tag to mark our recent release.
What you say about bugs is probably right.
The last one I found quickly ;-)



ps:

Total code changes is about 30000 lines in the last 3 month.
To compare: our mails without headers have about 40000 lines in the
last 9 Month.  So we roughly code 2.25 times as much as we write mails.
(I excluded mails with attached files.)

Mail lines stats:
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max.  Variance  Std Dev.
   8.00   20.00   28.00   39.52   46.75  467.00  1339.517  36.59941




-- 
Kai Antweiler