From: Vladimir Prus (ghost_at_[hidden])
Date: 2003-01-05 10:33:54
David Abrahams wrote:
>>> will be used to denote generator, to which flags apply.
>>>What does it mean that flags will apply to a generator? Is the idea
>>>simply to limit the "on target" variable assignments to the products
>>>of that generator?
>>Kind of. Generator id has a second function. It names a rule which is
>>used to actually set up build action. E.g. gcc.link generator has
>>associated gcc.link rule. (In fact, gcc.link *rule* is empty, there's
>>only gcc.link *action*, but that's detail).
>>I meant that if you say that flags apply to gcc.link, that the variable
>>assingment will be only done for that rule, but not for gcc.compile, for example.
> What do you mean by saying a variable assignment will be done for a
> I can't change the contents of the rule to add a variable assignment,
> and we don't want to burden action writers with explicitly adding some
> boilerplate code to do it. It seems possible to have the generator
> base class perform the variable assignment before the generator
> invokes the rule, which is what I was saying, and seems to correspond
> to what you're saying below.
Your iterperation is correct. This is what is done at the moment. Seach
for the call to "set-target-variables" in virtual-target.jam to see
the logigic. (It's at line 521 in my copy).
>>>Also, I don't know how to reach this link mentioned in BB4 anymore:
>>Hmm... I get to it simply by clicking in Mozilla... Can Outlook open
>>news article by URL?
> I don't use Outlook.
> Outlook Express says first,
> "The server could not be found"
> and second:
> "Outlook express could not download the requested message. It is
> likely that the message was removed or expired from the server"
> I assume it's this one, though:
In fact I meant
but it's the same thread anyway. I wish there were an automatic
method to obtain message id that everybody can use later, but
nntp:// URLS doesn't work either.
>>P.S. BTW, there's now a long row of holidays in Russia. I had a week of them
>>already, and there will be two more: 6th and 7th. I'll be online as usual
>>after that (and will resume Boost.Build hacking as well).
Eer... of course, I can be buried under miriad of error messages from
BGL-using code ;-)
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