Super long compile times

3 views
Skip to first unread message

Peter Keen

unread,
Mar 1, 2010, 12:51:15 AM3/1/10
to ledge...@googlegroups.com
Is it normal for the master branch of ledger to take hours to compile?
It's going on two and a half right now in the middle of acprep opt
make. I would attach the actual output but my system is just sitting
there mostly locked up compiling. Standard snow leopard MacBook c2d.

The targets that seem to be taking forever look like this:

CXX libledger_math_la-amount.lo

Any ideas?

Thanks,
Pete

John Wiegley

unread,
Mar 1, 2010, 1:07:44 AM3/1/10
to ledge...@googlegroups.com
On Mar 1, 2010, at 12:51 AM, Peter Keen wrote:

> Is it normal for the master branch of ledger to take hours to compile?
> It's going on two and a half right now in the middle of acprep opt
> make. I would attach the actual output but my system is just sitting
> there mostly locked up compiling. Standard snow leopard MacBook c2d.

It all depends on your machine and what else it's doing. On my MacBook Pro, with 3G of RAM, I can run this:

acprep -j2 out update

But only if I'm not doing anything else with the machine. And yes, optimized builds can take quite some time, but 2.5 hours? That seems a little crazy. I'd expect 20 mins at most.

John

Peter Keen

unread,
Mar 1, 2010, 1:16:58 AM3/1/10
to ledge...@googlegroups.com
On Sun, Feb 28, 2010 at 10:07 PM, John Wiegley <jwie...@gmail.com> wrote:
> It all depends on your machine and what else it's doing.  On my MacBook Pro, with 3G of RAM, I can run this:
>
>  acprep -j2 out update
>
> But only if I'm not doing anything else with the machine.  And yes, optimized builds can take quite some time, but 2.5 hours?  That seems a little crazy.  I'd expect 20 mins at most.

My MacBook is a C2D 2.15GHz with 512mb and only running Chrome, emacs
with two buffers open, and iterm with one terminal going. I'm
wondering if it had something to do with the homebrew formula I was
using, because it's going a little bit faster now that I'm not using
that. I'm going to let it run overnight tonight and hope it's done by
morning. I'll attach the log tomorrow.

John Wiegley

unread,
Mar 1, 2010, 1:18:34 AM3/1/10
to ledge...@googlegroups.com

Ah, it's the 512mb that's killing you. Check your Activity Monitor, I'm pretty sure your machine is swapping a lot.

John

Peter Keen

unread,
Mar 1, 2010, 1:24:09 AM3/1/10
to ledge...@googlegroups.com
On Sun, Feb 28, 2010 at 10:18 PM, John Wiegley <jwie...@gmail.com> wrote:
> Ah, it's the 512mb that's killing you.  Check your Activity Monitor, I'm pretty sure your machine is swapping a lot.

Ahhhhh, yeah there it is. Dang it. At some point in the process of
going from 2.6.2 to 3.0 did you happen to add --fuse-all-the-ram?
Would it be safe to remove that? ;)

--Pete

John Wiegley

unread,
Mar 1, 2010, 1:43:04 AM3/1/10
to ledge...@googlegroups.com

It's the use of Boost, and all its templates -- plus the optimization flag -- that's doing it. :(

John

Peter Keen

unread,
Mar 1, 2010, 2:21:57 AM3/1/10
to ledge...@googlegroups.com

No worries. I added -j1 and now it actually gets all the way to link
in a reasonable amount of time! Link is failing, but that's a quirk of
my setup. Libtool thinks that macports still exists when in fact it
does not. I'll figure that out and then I should be up and running.

Anyway, boost is great. We use it all the time at work and I wouldn't
do a large c++ project without it.

--Pete

Reply all
Reply to author
Forward
0 new messages