From: Ali Azarbayejani (ali_at_[hidden])
Date: 2003-03-20 16:34:27
David Abrahams wrote:
> Vladimir Prus <ghost_at_[hidden]> writes:
> > I'm very interested in do-nothing timing. Can you run "bjam -d+10" on an
> > unmodified tree and send me the output. This will tell what consumes most
> > time.
> >> <B> Command line overflow on link (Show-stopper!). Some reasons
> >> <C> Command line overflow on compile (Show-stopper!). Some reasons
> >> <D> Command line overflow on archive/dll (Show-stopper!). Some
> Something's suspicious about this, at least if he's using the gcc
> toolset... Oh, yeah; I forgot we need to add a small enhancement for
> potentially long one-line actions on NT.
> I'm kind of shocked about the problem on compilation though.
> Ali, are you using Win98 or something?
No, I'm using Linux (RedHat 7.2)
The problem on compilation was due to a gazillion "-I<path>" arguments,
many of which were duplicated many times and each of which was supremely
verbose because of absolute paths deep into the file system. Vladimir
checked in a change early in the week that solved the overflow problem
for compile, leaving me with overflow on archive/dll (but only as a
sub-project, when long absolute pathnames are involved) and with final
link (which wouldn't occur if the libraries were not duplicated).
The solution to all the remaining current problems was recompiling bjam
with jam.h:MAXLINE=100000 (instead of 10240) as proposed by Jurgen.
However, I think it is still worth reducing the command line verbosity
where possible, as it will bite sooner or later, and makes command lines
a bit easier to look at.
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