From: Jeff Garland (jeff_at_[hidden])
Date: 2001-10-04 11:06:18
> On one hand, we should certainly encourage commercial support for Boost
> libraries, and one really good way to do that is via links on the Boost web
> site. Perhaps on a separate "Support" page.
Agreed. Strangely, some companies will not use a library unless they can pay
someone for 'support'. This is especially true if they have a 24X7 operation
and they don't want to maintain in-house expertise. Sometimes it is simply
strange corporate policy.
> On the other hand, how do we prevent incompetents or even outright crooks
> from submitting links for a Boost Support page?
I don't think you need to. First off, you will need to have more people come
forward to offer support -- that might not happen. That said, if a company is
going to pay for support it is their responsibility to evaluate the company they
are buying from. It might be more useful to provide a list of suggested
criteria for evaluation of capability. For example, a query of mail archives
would give insight into level of past participation in boost libraries. Of
course, if the support contract is limited to smart_ptr and the participation is
in the development of regex, what does this mean? My point is that there are
too many unique factors for each situation. Also, if you restrict the listings,
you will be implicitly deciding on winners and losers.
> (Obviously we would have a disclaimer on the web page, but I'd personally
> feel more comfortable with active filtering by several people.)
> Perhaps require that the links be to persons or organizations known to be
> familiar with Boost.
> If so, "known" to who? The moderators? A special group of Boosters? Any
> Boost member?
> And what constitutes being "familiar with Boost"? Being a long-time
> member? A past message poster? A Boost developer? Any of those?
> What about a company or consultant already active in the support business,
> but with no history of Boost evolvement? Personally, I'd rather start with
> tight limits requiring past involvement, then loosen up later if indicated.
Maybe the support page should simply suggest using google to search for "Boost
C++ Support" or some other phrase (along with suggested evaluation criteria).
Companies that wish to provide support should have a web-site and should be
easily found via this mechanism. This circumvents the issue of endorsement. It
makes it absolutely clear that their is no evaluation by Boosters of the support
Just to take the other side of all this for a moment, Doug Schmidt has taken the
exact opposite approach with ACE-TAO. He has formed support alliances that he
endorses specifically. The big difference is that ACE-TAO has a single
technical leader while Boost is much more a group collaboration.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk