Boost logo

Boost :

Subject: Re: [boost] [local] Any "active Boost library author" in favor of Boost.Local?
From: Andrzej Krzemienski (akrzemi1_at_[hidden])
Date: 2011-11-25 04:23:57


> Hello all,

> I will be honest: The fact that some of the people arguing for not
> including Boost.Local are "active Boost library authors" (Thomas,
> Joel, and Hartmut) seems to make their opinion weight more than the
> opinion of others that have also submitted a review for the library.

> Two questions:

> 1) Should reviews from "active Boost library authors" carry more
> weight? (This is /not/ a rhetoric question.)

Hi Lorenzo.

I am /not/ an active Boost library author or maintainer. I will only
address the first question. I was asking myself the same question recently,
and my answer to it is that their opinions should weigh more than others'.
Well, "others" is an ambiguous term; I meant someone like me, who is just a
Boost user, and says "I want to have this, and this, and that one". My
argument is that Boost as an organization has its goals or status, or
spirit (no pun intended). I tried to look for something like "what is our
goal" on the Boost site, but I did not find a satisfactory answer. The only
thing that I found useful is a section in Scott Meyers's "Effective STL".
I am not sure if Boost developers will agree, but he says that Boost was
founded in part by people who submitted proposals for Standard Library to
the C++ committee that were rejected. Boost was supposed to be the place
where these libraries were still to be developed, maintained and widely
used, which would give them the grater chance to succeed in the next round
of standardization.

On the other hand, some existing Boost libraries do not appear (at least in
my limited view) to meet this criterion (as well as other criteria
mentioned recently). I am thinking about the unit testing framework. Do not
get me wrong I am not trying to imply that it shouldn't be there, and in
fact this is my preferred unit testing framework for many valid reasons.
But I was trying to apply the recently risen acceptance criteria to all
Boost libraries that I know, and this example appeared the most
interesting. Even though it provides a non-macro alternative it encourages
using macros for nearly everything in the "automated version": tests,
checks, suites, modules. Even thought there are other free libraries for
C++ that are known to provide almost the same functionality without a
single macro: http://tut-framework.sourceforge.net/. It does not look like
a potential candidate for the inclusion into the standard; there are many
unit-testing frameworks out there that do not appear to be any worse. It
does not appear to encourage some good coding style (apart from the fact
that having good unit tests is a good development style)....

Given this lack of clear criteria, active Boost library authors and
maintainers should be in the better position to assess if the tacit
criteria for library admission are met. Some of them have been with boost
since its inception (almost 15 years now). If we think otherwise, we are
risking the situation where Boost users that just want some library "in the
Boost package" will outnumber the few library developers that have a
broader goal in mind, and may not even have time to submit a review because
they are busy maintaining their libraries or submitting proposals to ISO
Committee.

I guess that in case of Boost.Local review, this argument would be stronger
if the "active Boost library author" reviewers were not the authors of
libraries that /somehow/ overlap with Boost.Local. Thus I believe you are
bringing the valid point in trying to engage other Boost library authors,
and it would be beneficial, also to the entire Boost community, if someone
responded to your question.

Regards,

&rzej


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk