Re: [Rd] SVN-REVSION altered when building R-devel out of tree from last snapshot

From: Martin Maechler <>
Date: Tue 20 Dec 2005 - 08:40:47 GMT

>>>>> "Herve" == Herve Pages <>
>>>>> on Mon, 19 Dec 2005 17:10:58 -0800 writes:

    Herve> Hi,
    Herve> Today I downloaded and compiled the last R-devel snapshot.
    Herve> The SVN-REVISION in the tarball contains the following:

    Herve> Revision: 36792
    Herve> Last Changed Date: 2005-12-18

    Herve> But after compiling on Unix (I compiled out of tree),

i.e. "in a separate build directory tree"

    Herve> I ended up with an SVN-REVSION file containing:

    Herve> Revision: unknown
    Herve> Last Changed Date: Today

    Herve> in the build tree.

I can confirm this wrong behavior (Linux Redhat EL4). There must be something not yet perfect in our 'make' setup there. If we are not in the srcdir, we create a 'non-tarball' file which I think is wrong; in any case, this is buglet we'll fix.

Thank you, Herve!

    Herve> Then when I start R, I get:

    Herve> R : Copyright Today, The R Foundation for Statistical Computing
    Herve> Version 2.3.0 Under development (unstable) (Today-Today-Today)
    Herve> ISBN 3-900051-07-0

    Herve> even if I naively edit the SVN-REVISION in the build tree before to     Herve> start R.

    Herve> I got this problem on a 64-bit SUSE Linux 9.2, a 32-bit SUSE Linux 9.2
    Herve> and a Solaris 2.9 sparc system.
    Herve> On Windows however (where I built R directly in the source tree) I don't
    Herve> have this problem.

    Herve> We need to update R-devel on our various build machines in order to test
    Herve> Bioconductor devel packages with last R-devel and we try to have the exact
    Herve> same R revision number on every test-machine. Last time I updated R-devel     Herve> was 12/01/2005 and I used the same procedure that I

[ you mean 12th of January? ;-) {yes, it would help to use   international 2005-12-01 or then Dec 01, 2005} ]

    Herve> am using today but
    Herve> I didn't have the SVN-REVISION problem.

    Herve> Also I didn't try to build R-devel from SVN. Maybe     Herve> this could solve the problem.

that would definitely solve it, since that's what all of R-core do "all the time".
But the way you did, should also work; that's what the tarballs are for!

    Herve> It's just that using the tarball was easier to manage.     Herve> Anyway I thought it might be worth reporting.

Thank you again, Hervé !

    Herve> Regards,

    Herve> Hervé  

    Herve> -- 
    Herve> ------------------------
    Herve> Hervé Pagès
    Herve> E-mail:
    Herve> Phone: (206) 667-5791
    Herve> Fax: (206) 667-1319

    Herve> ______________________________________________
    Herve> mailing list     Herve> mailing list Received on Tue Dec 20 19:55:56 2005

This archive was generated by hypermail 2.1.8 : Mon 20 Feb 2006 - 03:21:35 GMT