Re: [Bug-gnubg] Three things that are broken

[Jim Segrave]

On Sat 14 Apr 2007 (10:22 +0000), Jonathan Kinsey wrote:
> 
> >From: Christian Anthon <address@hidden>
> >
> >- Python support
> >a) It is buggy
> >b) Nobody, even among the developers use it regularly.
> >c) It will never be used by a very large fraction of the users.

I don't know how it is broken, under Linux and FreeBSD it works. I use
it regularly and I think that it should be the first choice for adding
features to gnubg, rather than introducing new C code. I fail to see
how removing it and finding a new way of doing rdbms support would be
anything but a retrograde step.

> >- Time controls
> >a) it doesn't work out of the box
> >b) it crashes (at least for me)
> >c) It will never be used by a very large fraction of the users.

No opinion, I have never used it

> >- Player records
> >a) the player records are under featured
> >b) the relational database requires some setup and is overkill in its
> >construction
> >c) It is used by a large fraction of our users
> 
> It was always my understanding that the old player record code would be 
> removed when the relational database was working.  It's an important 
> feature so we should try and finish off the new player record database 
> (with some nice reports and graphs and things...).

I don't use the old player records, I do use the rdbms features
extensively and regularly. It would be good to make accessing the
database from within gnubg easier. If you're going to do this, then
retaining Python support is a must. You really don't want to start
writing database query support and stats processing as an integral
piece of gnubg code.

> >So what do you think? My personal preference is that we terminate python
> >support and time controls. And I prefer the relational database, but it
> >needs some adjustments, especially if we get rid of python support.

Vehemently opposed to removing python support. This is a major
mistake. If it's broken on some systems, lets get it fixed. One of my
projects which I'm finding difficulty to find free time for but really
would like to finish is to add the option to put errors into the
database and allow players to select a previous error at random and
replay it. I think this would make a very effective training
tool. Doing this in C code would be possible, but very
difficult. Doing this using Python requires adding a small number of
additional Python functions into gnubg.

-- 
Jim Segrave           address@hidden