tech-pkg archive

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

Testing NetBSD Versions



    Date:        Wed, 17 Dec 2008 00:42:42 +0100
    From:        Adam Hoka <adam.hoka%gmail.com@localhost>
    Message-ID:  <20081217004242.116ee257.adam.hoka%gmail.com@localhost>

This particular message just recalled to mind something I had been
meaning to mention for a while, I mean no specific comment about
Adam's message (or fix) in particular ...

  | diff -u -r1.51 Makefile
  | --- Makefile        4 Aug 2008 06:55:39 -0000       1.51
  | +++ Makefile        16 Dec 2008 23:41:38 -0000
  | @@ -24,7 +24,7 @@
  |
  |  GCC_REQD+=         3.0
  |
  | -.if !empty(MACHINE_PLATFORM:MNetBSD-4*)
  | +.if !empty(MACHINE_PLATFORM:MNetBSD-[4-9]*)
  |  USE_PKGSRC_GCC=3D  yes
  |  .endif

This kind of thing appears quite often in pkgsrc - and while it isn't urgent
that it get fixed in a hurry, someone really should be planning the correct
invocation to handle NetBSD 10.0 and beyond (both for testing the "must have
at least" case, and the as here "want to know if >= than" case).

Otherwise, when NetBSD 10 does appear (7-8 years ??) there will be a lot
of stuff that suddenly breaks (just like qemu's pkgsrc did here for NetBSD 5).

kre

ps: and this is aside from the issue that MACHINE_PLATFORM almost always
says what kernel is running, which doesn't necessarily have a lot to do
with what userland is installed, and the latter is what is really needed here.

Even more someone might easily have updated their gcc on a NetBSD 3 system
to gcc 4 for example, in this particular case, the correct test would be
whether the system's gcc is gcc 4 or later, not whether a NetBSD 4 or later
is installed - and the same the other and probably more common way, I have
systems with NetBSD 4 (or even 5_BETA) kernels installed that are still
running NetBSD 3 userland (maybe one that is still NetBSD 2 userland).
In that case here, there is no real problem, just a possible unnecessary
install of pkgsrc's gcc3, but in others bad things happen.   Testing system
version numbers is rarely the right option, it's just the easy way...



Home | Main Index | Thread Index | Old Index