Re: Boost.Threads draft library submission

From: williamkempf_at_[hidden]
Date: 2001-03-15 12:12:37

--- In boost_at_y..., Beman Dawes <bdawes_at_a...> wrote:
> Bill Kempf wrote:
> 
>  >... I've uploaded a _DRAFT_ library submission for
>  >Boost.Threads.
> 
> Great!  Here are some initial reactions:
> 
> *  The headers should go in a boost/threads sub-directory.  We 
should do 
> this will all boost libraries with more than one or two headers, 
> particularly ones like Boost.Threads which will probably grow in 
terms of 
> number of headers.  Relatively easy to do now, messy to do later.

Ok, I'll make this change in the source and documentation.

> *  Your documentation deserves compliments.  It is easy to read, 
with 
> reasonably sized paragraphs and sections, and short, clear, easy to 
> understand sentences.  I hope as it inevitably expands, you will do 
so by 
> adding additional sections and pages, and keeping the current to-
the-point 
> style.

Thanks.  Some expansion should, and will, go directly into the 
existing pages, but I'll be careful to keep things short and easy to 
comprehend.
 
> *  Several other folks (Jeremy Siek and Dietmar Kuehl come to mind) 
have 
> written papers regarding Boost threading issues.  Should these be 
included 
> in the docs package?  (I haven't looked at them in awhile, so don't 
know if 
> they still apply.)

Most of what's been written has been included in the documentation 
(reworded to fit) if it's applied to the current state of the 
library.  What's missing from the draft is an aknowledgements page 
citing these contributions.  This wasn't an oversight, or meant as a 
slight.  At least one of these folks helped enough that I'm offering 
them inclusion in the copyright.  The rest deserve recognition and 
I'll be working on an aknowledgements page, but this will take a bit 
of time and effort so I've not done so in the draft.  It *WILL* be 
included before formal submission.
 
> *  You might want to consider starting a FAQ fairly soon.  While 
some of 
> the questions coming up can be addressed by improving the core 
docs, some 
> are more FAQ-like and should be treated as such.

Good idea.  I'll look into this... though until we reach 
some "critical mass" of such FAQs it may not be worth publishing.  So 
for now I'll just keep a list privately.
 
> *  Skim reading your interfaces, it is really nice that you have 
held the 
> number of functions down to a reasonable number. That will make 
inspection, 
> testing, documentation, understanding, and a lot of other things 
easier.

Thanks, though several others deserve some credit on the final 
interface designs as well.
 
> Finally, thanks for all the effort you've put into Boost.Threads!

I'm just hopefull that Boost can insure that my efforts result in a 
truly usable implementation.  I know that during the design phase I 
was a little head strong in some of my opinions, but now that I've 
got a working draft I plan to be a LOT more open to ideas and 
concerns, so I'd like to encourage people to re-voice any that I 
dismissed in the past that they still feel strongly about.

Bill Kempf