Re: [jamboost] Semaphores

From: Rene Rivera (grafik666_at_[hidden])
Date: 2002-11-22 01:20:04

[2002-11-21] Vladimir Prus wrote:

>
>I commonly use "-j 4" flag to bjam, because it's possible to
>distribute compilation to four machines. Unfortunately, linking
>is not distributed, and this seems affecting performance ---
>those four linkers are all accessing the same drive.
>
>To overcome this, I plan to adopt the extension described in
>
>http://maillist.perforce.com/pipermail/jamming/2002-April/001667.html
>
>which allows to serialize executing of certain build actions.
>
>Are there any objections?

I don't see how this can solve your problem.

How is it that the linking is not distributed?

>If not, I've a question to Rene. I need to skip targets for which the
>corresponding semaphore is locked. I.e. instead of extracting it from
>stack and processing, I'd like to put it to the end of stack. Which way is
>best? Should I simply store a pointer to the last state?

Here I don't think that will work. If you put it at the end of the stack
won't that make it break the build dependency order?

-- I'm a little foggy on the stack code... as I didn't write it... Steve
Anichini did. Steve you still paying attention to this list?

-- grafik - Don't Assume Anything
-- rrivera_at_[hidden] - grafik_at_[hidden]
-- 102708583_at_icq - Grafik666_at_AIM - Grafik_at_[hidden]