I just pushed the first example of a configure script

David Holmes David.Holmes at oracle.com
Thu Aug 25 05:15:42 PDT 2011


On 25/08/2011 8:53 PM, neugens.limasoftware at gmail.com wrote:
> I apology for top posting, I really hate that, but my phone doesn't support
> anything else...
>
> Yes, the build directory is the place you generate the configure file and
> where the resulting binaries are generated, and by default the sources are
> searched either in the current directory or in the parent one if I remember
> correctly.
>
> You don't need to rerun configure all the time, only if you change
> configuration details, as make takes care of the build itself.

But my output directory is often on a different file system (local rather 
than NFS) and it is determined by build identifier eg:

b00/linux-x
b00/linux-y

etc.

and then

b01/linux-x
b01/linux-y

etc.

So I'm unclear where exactly the configure directory exists in relation to 
this output structure.

> If I understand correctly your question, that means to run configure one
> time per each build target, but only once (unless you change configuration
> details, like adding a new library as a dependency for example).
>
> Do you now only run the build setup once for all the targets?

What build setup do you mean? The only thing I change in my build script is 
the current build identifier. Otherwise I just invoke scripts that set the 
necessary environment variables for the target platform (whether local build 
or cross-compilation build).

Cheers,
David

> I think the only difference from now would be the extra configure step.
>
> Cheers,
> Mario
> --
> Sent from HTC Desire...
>
> pgp key: http://subkeys.pgp.net/ PGP Key ID: 80F240CF
> Fingerprint: BA39 9666 94EC 8B73 27FA FC7C 4086 63E3 80F2 40CF
>
> http://www.icedrobot.org
>
> Proud GNU Classpath developer: http://www.classpath.org/
> Read About us at: http://planet.classpath.org
> OpenJDK: http://openjdk.java.net/projects/caciocavallo/
>
> Please, support open standards:
> http://endsoftpatents.org/
>
>
> ----- Reply message -----
> Da: "David Holmes" <David.Holmes at oracle.com>
> Data: gio, ago 25, 2011 12:36
> Oggetto: I just pushed the first example of a configure script
> A: "neugens.limasoftware at gmail.com" <neugens.limasoftware at gmail.com>
> Cc: "Magnus Ihse Bursie" <magnus.ihse.bursie at oracle.com>,
> <build-infra-dev at openjdk.java.net>
>
>
> On 25/08/2011 8:30 PM, neugens.limasoftware at gmail.com wrote:
>  > What I mean is that the build directories can still be grouped, and that
>  > configure should allow the use of the --srcdir and similar options to allow
>  > such flexibility.
>
> Is your build directory the place where configure generated its output and
> the place in which your build output is placed?
>
>  > It's true that you still need to run configure from the given build dir
>  > though, but this can be automated via scripts as well.
>
> I don't want to re-run configure each time I change the build identifier.
>
> I think I need to see a concrete example of how the src code, build files
> and the configure output are organized to understand how I would have to
> adapt it to suit my needs.
>
> Cheers,
> David
>
>  > The idea would be to make cross compilation easier even on those platforms
>  > that are not officially supported yet I think.
>  >
>  > Mario
>  > --
>  > Sent from HTC Desire...
>  >
>  > pgp key: http://subkeys.pgp.net/ PGP Key ID: 80F240CF
>  > Fingerprint: BA39 9666 94EC 8B73 27FA FC7C 4086 63E3 80F2 40CF
>  >
>  > http://www.icedrobot.org
>  >
>  > Proud GNU Classpath developer: http://www.classpath.org/
>  > Read About us at: http://planet.classpath.org
>  > OpenJDK: http://openjdk.java.net/projects/caciocavallo/
>  >
>  > Please, support open standards:
>  > http://endsoftpatents.org/
>  >
>  >
>  > ----- Reply message -----
>  > Da: "David Holmes" <David.Holmes at oracle.com>
>  > Data: gio, ago 25, 2011 12:16
>  > Oggetto: I just pushed the first example of a configure script
>  > A: "neugens.limasoftware at gmail.com" <neugens.limasoftware at gmail.com>
>  > Cc: "Magnus Ihse Bursie" <magnus.ihse.bursie at oracle.com>,
>  > <build-infra-dev at openjdk.java.net>
>  >
>  >
>  > On 25/08/2011 8:08 PM, neugens.limasoftware at gmail.com wrote:
>  > > You can add this to your script ;)
>  > >
>  > > Seriously, I'm not sure to follow the problem, but I think you cannot mix
>  > > the build directories for different targets anyway.
>  >
>  > What do you mean by a "build directory"? The output directories are distinct
>  > based on platform, but grouped in a common location and arranged by a build
>  > identifier.
>  >
>  > David
>  >
>  > > I would expect that configure takes an argument to identify where you want
>  > > the build to take place as well, rather than use an hardcoded default, so
>  > > simply adapting the build script would do the same, really.
>  > >
>  > > Cheers,
>  > > Mario
>  > > --
>  > > Sent from HTC Desire...
>  > >
>  > > pgp key: http://subkeys.pgp.net/ PGP Key ID: 80F240CF
>  > > Fingerprint: BA39 9666 94EC 8B73 27FA FC7C 4086 63E3 80F2 40CF
>  > >
>  > > http://www.icedrobot.org
>  > >
>  > > Proud GNU Classpath developer: http://www.classpath.org/
>  > > Read About us at: http://planet.classpath.org
>  > > OpenJDK: http://openjdk.java.net/projects/caciocavallo/
>  > >
>  > > Please, support open standards:
>  > > http://endsoftpatents.org/
>  > >
>  > >
>  > > ----- Reply message -----
>  > > Da: "David Holmes" <David.Holmes at oracle.com>
>  > > Data: gio, ago 25, 2011 12:01
>  > > Oggetto: I just pushed the first example of a configure script
>  > > A: "Magnus Ihse Bursie" <magnus.ihse.bursie at oracle.com>
>  > > Cc: <build-infra-dev at openjdk.java.net>
>  > >
>  > >
>  > > Hi Magnus,
>  > >
>  > > On 25/08/2011 7:35 PM, Magnus Ihse Bursie wrote:
>  > > > Hi David and everyone else,
>  > > >
>  > > > I'm an Oracle engineer that have just barely returned from my paternity
>  > > > leave, and I will be helping Fredrik, Erik and Kelly with the build
>  > > > infrastructure project, as soon as I get up to speed. :)
>  > > >
>  > > > On 2010-10-22 02:43, David Holmes wrote:
>  > > >> I guess I'll have to reserve judgment but I don't like the idea of
> having
>  > > >> to setup different build directories in which to run configure for local
>  > > >> vs cross-compile - so far that's not making my job (as a developer) any
>  > > >> easier.
>  > > >
>  > > > The goal of this project is of course to make the everyday work situation
>  > > > for the developers better - not worse!
>  > > >
>  > > > Can you please elaborate on what the problem is? If I understand you
>  > > > correctly, you have a situation where you want to repeatedly build
> the same
>  > > > source code to the local platform and also to one (or more) foreign
>  > > > platform(s) using cross-compile, right?
>  > >
>  > > Right.
>  > >
>  > > > How would you accomplish this today and how does it differ from using
>  > > > configure?
>  > >
>  > > Today I have a source forest and a top-level builds directory that contains
>  > > my build scripts for the different platforms. I simply do build-x.csh or
>  > > build-y.csh.
>  > >
>  > > My recollection from building configure based tools is that now I would
> have
>  > > to do something like:
>  > >
>  > > mkdir build-x
>  > > cd build-x
>  > > <run configure for platform x>
>  > > cd ..
>  > > mkdir build-y
>  > > cd build-y
>  > > <run configure for platform y>
>  > >
>  > > And thereafter I'd have to identify the right build-x or build-y directory
>  > > to do the actual build.
>  > >
>  > > I'm not saying this is intolerable just saying that this is more cumbersome
>  > > than what I have to do now an
>



More information about the build-infra-dev mailing list