[DGD] DGD and *dbm

[bart at wotf.org]

Felix,

Maybe I worded things badly in my last post, it seems you are reading things
in it which I didn't intend to say.

On Wed, 23 Jul 2008 22:54:48 +0200, Felix A. Croes wrote
> 
> I'm not the one who turned this a professional vs. hobbyist issue. 

Nope, and it isn't a professional vs hobbyist issue. It is however true that
both have different and at times conflicting points of view, and I don't think
there is anything bad about realizing that.

Beyond this, hobbyists get a very well maintained and supported driver, so I
wasn't arguing that 'you are leaving us hobbyists out' or anything like that,
that would be ridiculous.

>  As far as I'm concerned, all that matters is that the software is designed
> and written by me.  I'll gladly listen to suggestions, but if what
> you propose is in my view an inferior solution, you are simply out of
> luck.

Its definitely yours, and if you don't want something in there, then thats it.

> Badgering me over and over about the same issues will accomplish nothing.

You asked me if I have plans for DGD/MP for the *dbm package I am considering,
I pointed out why I can't right now, and Shentino pointed out that I possibly
could if an api existed for doing this kind of thing in a portable way.

I'm sorry but that seems a reasonable comment to me.

> Neither will the insinuation that I owe something to people less
> fortunate than myself,

If that is what you read there, I'm sorry, no, you do not owe such things to
anyone, and I didn't mean to suggest that.

What I did mean is that having such an interface will enable people to write
extensions for the areas THEY are good at, without having to be good at
cutting into the internals of DGD. That is something that could make DGD more
attractive to use for more people.

> or appeals to fairness.

My comment about fairness concerned your critisism of Shentino, not wether you
implement some feature or not. The later is entirely and solely your choice. 

> In my view, rejecting bad ideas is one of my most important functions.

Absolutely, and I trust you to do a rather good job at that. I do however
disagree about your judgement at times. When I can't convince you and still
think a feature is a good idea, I'll go try and implement it anyway, as might
be obvious from what I distribute.

Doing the work is not the issue here, I'm quite prepared to do the work
required. However, when it comes to an improved extension interface there are
a few reasons why I can't:

1. It has to provide a path to migrate extensions to DGD/MP
2. It has to be part of a mainstream DGD distribution.

If those 2 aren't matched at least, such an interface will have very limited
usage. To me this means that it either gets implemented by the author of the
software, get implemented based on a design by that author, or not at all.

If I end up creating the dbm package, I'll surely end up with some common
interfaces for it and the networking package, which will go a bit into the
direction of an extension interface, but it can't hope to provide any
migration path to any version other then the current sp development version.

Saying that people who argue that an improved extension interface is a good
idea are trying to push *THEIR* work onto you is what I called unfair, it is
not *THEIR* work. Its not yours either unless you happen to want to do it.

> 
> Regarding an improved extension interface, you are free to make
> suggestions.  As I have nothing further to add beyond what I wrote
> on this subject half a year ago, I will let the matter rest there.

I'll keep that in mind once the time is there.

As for a previous suggestion, using arrays of arrays for a dbm index, looking
a bit closer at concurency issues suggests that is the better solution, it
prevents a whole lot of problems with keys being added/removed

Bart
--
Created with Open WebMail at http://www.bartsplace.net/
Read my weblog at http://soapbox.bartsplace.net/