From: Vladimir Prus (ghost_at_[hidden])
Date: 2002-06-28 04:51:14
David Abrahams wrote:
> > > > Vladimir, what do you think needs to happen next? Did you guys agree
> > > > with
> > > > my radical command-line suggestions? I could implement that stuff...
> > >
> > > Hi David,
> > > I don't think your suggestions are all that radical! I think that it's
> > > very
> > > reasonable and it would be great if you implement it.
> > David, have you made any progress on this?
> I haven't started, sorry.
> > If it is not close to finishing,
> > then probably we can change tasks a little. I think I understand how it
> > should be implemented.
> Yeah, well it's pretty straightforward.
> > On the other hand, you had some ideas about generating
> > paths from proprty sets and have implemented 'feature.jam', so could
> > write impoved property.as-path rule better and faster than I. Would you
> > mind to implementing that in the first place, leaving command line to me?
> No problem at all to switch tasks, but I'm not sure what the improvement is
> that you're referring to. The only mention I can find of "as-path" in the
> message archives is the one I'm replying to.
> Maybe you mean item 1. in
> http://groups.yahoo.com/group/jamboost/message/1205? Or did you have
> something else in mind?
I'm referring to issue 553727 in the tracker. The message you mention applies
as well. The property.as-path rule is used now to generate subvariant paths
that correspond to property sets. It does so in a very dump fashion, just
creating a new path element for each property. So we'll have very funny paths:
You see, debug is implicit property and "variant-" is not needed.
Then, <optimization>ofis part of "debug" an should be omitted also.
IOW property.as-path should generate the most compact representation
of the properties given to it.
> P.S. I realize that I'm very literal-minded and that may sometimes make
> things difficult...
Hmm... I really don't understand why liberal mind causes any problems? I even
looked up the word in the dictionary but it still means what I though.
> it would help if you could check postings for possible
> ambiguities - if we were in the same timezone it would probably not make so
> much difference, but in our case each miscommunication tends to cause a
> 24-hour delay (presuming we get it cleared up in one round).
Okay, I'll try to put more effort in it.
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk