[Libreoffice] [bjoern.michaelsen at canonical.com: Re: make experts: progress information?]

Bjoern Michaelsen bjoern.michaelsen at canonical.com
Thu Oct 20 01:45:30 PDT 2011

On Thu, Oct 20, 2011 at 09:13:20AM +0100, Michael Meeks wrote:
> 	Well - the problem with this is that (by design) tail_build is a single
> make instance that compiles ~all of LibreOffice :-) As you say, that
> does break the usefulness of the zenity output - but it also makes it
> really rather hard to extract the progress information I'm afraid.

Well, it would be quite simple to call a make method on every compiled object
which increases a global counter. However, usually one is interested in the
fraction of the build that is completed and not in the absolute number. 
For that, one would either need to:
* calculate the total number of outdated targets (something which could be done
  with an extra "make -p" run, but that slows things down)
* just use the number of total declared compiled objects, however that progress
  would look odd for incremental builds as up-to-date targets are not

So the first method would be a slowdown esp. for incremental build (Bad). And
the second one would be look odd on an incremental build: Assuming you had to
recompile 10% of the build it would creep slowly to 10% and then jump to 100%
suddenly. I could hack out both methods pretty easy, but am reluctant because
of the drawbacks.



P.S.: Sorry, resent to the list. /me is still training on mutt.

More information about the LibreOffice mailing list