PE>> Well I should add to that, "release it even if it isn't into a
 PE>> perfectly desirable state, people can add to it".

 RS> To some extent. OTOH PVert is a classic example of what IMO is better
 RS> than nothing, but it would have been better with some decent atleast
 RS> usage notes, since otherwise everyone who wants to try it has
 RS> effectively got to fart around with the code to produce informal usage
 RS> notes for themselves. Thats not a very efficient way to operate IMO.
 RS> Clearly its better than nothing at all, but many dont think its a
 RS> terrific idea to be releasing stuff as minimally as that.

I think PVERT is something which aids my case considerably.  It was
released with no docs whatsoever, but people are going to start using it,
and writing documentation for it.  Which is a hell of a lot better than
letting it sit on his hard disk because it isn't in a releasable form yet. 
Which is what you were saying people were doing ("not being
pricks").  I'd rather have something than nothing.  Which is basically
why I wanted Tobruk to be out in some form. At least it does *something*.

 PE>> Once some of these guys start using PKT2QWK (which I have read some of
 PE>> the code so that I can tell them what options to use), I'm sure I can
 PE>> bully them into writing up some docs for it.

 RS> Sure, thats always feasible. OTOH IMO it makes more sense for the author
 RS> of the code to have done that.

Of course.  It would also be more useful if the author sent a copy to
everyone in the world who wanted it, with a ring-binder manual.  I'm saying
PVERT is a step forward, not 3 steps forward.  I'm not asking for the whole
world all at once, I don't know why you seem to insist on it.

 RS> Even with that, there is more to it than that too. You get into the
 RS> classic problem with librarys and APIs that some are based on an elegant
 RS> and homogeneous architecture and are convenient to use. You can get away
 RS> with having an abortion if you have the muscle like Windows has, but not
 RS> with less critical stuff always.

I think you might have left a "not" out, somewhere here.

 PE>> I personally have put my money on an interface spec like ISO C has.
 PE>> Do you have any other suggestions on making code reusable?  I would
 PE>> also suggest it needs a standard error-handling strategy, like I have
 PE>> supplied with error.c.

 RS> Its a problem, which you dont even see addressed that well with say the
 RS> Windows and OS/2 APIs. Its taking the industry a long time to get that
 RS> stuff right and IMO they are nowhere near there yet.

Well they are with the C standard library.

 PE>>> The technology you provide can NEVER be destroyed, because it will be
 PE>>> spread to BBS's all over the world.

 RS>> And much of it disappears very quickly.
 RS> Sure, I should have made clearer that I meant disappear in the sense of
 RS> actually fulfill a useful function and get used. Not literally disappear.

Yeah, well if people would release public domain spreadsheets and other
applications instead of DOS utilities, it might help!

 RS> IMO a lot of the Whereis type utes are too, even when in PD form. Much
 RS> of the need for them has been replaced by stuff which does it much
 RS> better like Magellan. So while they sit on disks all over the place,
 RS> they arent IMO contributing much to forward progress in software anymore.

I use dir xxxx*.* /s to find stuff.  I used to have a whereis program for
DOS (still do actually), but don't have one for OS/2.  BFN.

Paul

--- GoldED/2 2.42.G1114