Boost logo

Boost :

From: Jared McIntyre (jmcintyre_at_[hidden])
Date: 2007-05-06 19:44:25


> - What is your evaluation of the design?

I like the general design. It is easy enough to use.

> - What is your evaluation of the implementation?

I'd really like to see the MAC address + time generator. I realize that this
might be a pain to do in a cross-platform manner, but it would make the
library more complete in my eyes, and more likely for me to recommend it
to my colleagues. Is there a way that the library could be engineered so that
we could provide our own generation implementation?

> - What is your evaluation of the documentation?

I don't like it's organization. I'd much rather see it start out with the
Rationale section followed by a simple example of the most likely used
scenario. I think this is what most users are expecting to see who haven't
used the library before and want to get started quickly. You have to weed
into the documentation a while before you run into the first generation
examples, which is what I think most people will be intereste in. If you
need to put the "Class guid synopsis" in the docs, I'd put it near the end
since it is more-or-less a dump of the uncommented header file.

> - What is your evaluation of the potential usefulness of the library?

A good, cross-platform, UUID library is extremely useful. Without the
guaruntee of uniqueness provided by MAC address + time, it isn't as useful
for my situation. We have data generated across a large number of
disconnected computers (I highly doubt there would see a duplicate, but in
my case, we can't take the risk).

> - Did you try to use the library? With what compiler? Did you have any
> problems?

Played around with it with Apple's version of gcc 4. Worked fine.

> - How much effort did you put into your evaluation? A glance? A quick
> reading? In-depth study?

I read through the docs pretty thuroughly, and the implementation.

> - Are you knowledgeable about the problem domain?

I've used several libraries for different purposes. I have developed a
crossplatform wrapper in the past for platform specific libraries, but have
not had to implement that actual generator before.

I have a hard time saying that I'd recommend it for the library or not. I
think it would cover a lot of people's needs, but does fall a little short in
cases where you have to be absolutely sure that the generated ID will be
unique across systems. That is unless I'm miss-understanding how these
are generated.


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