pkgsrc-Bugs archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

pkg/38134: make update erases ancestor packages



>Number:         38134
>Category:       pkg
>Synopsis:       make update erases ancestor packages
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    pkg-manager
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Mar 01 20:55:00 +0000 2008
>Originator:     David A. Holland <dholland%eecs.harvard.edu@localhost>
>Release:        NetBSD 4.99.54 (20080229) (pkgsrc 20080229)
>Organization:
>Environment:
System: NetBSD tanaqui 4.99.54 NetBSD 4.99.54 (TANAQUI) #21: Fri Feb 29 
12:31:31 EST 2008 dholland@tanaqui:/usr/src/sys/arch/i386/compile/TANAQUI i386
Architecture: i386
Machine: i386
>Description:

        If you do a 'make update' for a package that has deps that are
        already installed, it will sometimes erase those deps. I am
        not sure of the exact circumstances or exactly what's going
        on, but it looks like pkg_delete -R may be chasing deps in the
        wrong direction.

        I am using the latest pkg_install in pkgsrc.

>How-To-Repeat:

        Find a place where package P depends on package Q, nothing
        else uses Q, and Q is already installed.

        Then recompile P with "make && make update".

        This happened to me just now with P=xine-lib and Q=vcdimager-devel.
        It erased vcdimager-devel, libcdio, and libcddb, and so the
        install phase of make update, which does configure --recheck,
        tried to use bits of vcdimager and bombed. In other cases, the
        ancestor package will just disappear silently so things will
        fail at runtime.

        If you choose P and Q such that other things use Q, pkg_delete
        will attempt to remove it and fail, generating messages in the
        build log.

        If you do just "make update" it will (I suspect) erase them
        before checking deps, and then silently recompile them before
        continuing, so the problem is probably often masked.

        I am not sure if this happens all the time or is triggered
        only by some packages or some kinds of deps or what.

>Fix:
        No idea.



Home | Main Index | Thread Index | Old Index