![]()
From: Kent Landfield (kent@hypermail.org)
Date: Wed Nov 17 1999 - 15:45:31 CST
# On Tue, 16 Nov 1999, Kent Landfield wrote:
#
# > First is, the maketgz and FILES file are there for the administative
# > purposes of just a couple people. They are of no use to the average
# > hypermail user. Shouldn't tools like this, that are only used for
# > building official distributions of hypermail, be in a separate CVS
# > baseline such as one named "hypertools" ?
#
# If they were many, big and extensive I would agree. But now, by keeping them
# in the main distribution archive we allow anyone at any time to build a new
# archive very easy. You could even make re-package hypermail again after you
# have untared it and thrown away the original tarball and it would become an
# exact copy.
I currently have a couple others. There should be a file that keeps all
the baseline tags and a description of what they are for. I understand
what you are trying to accomplish but it's just as easy to tar the whole
directory and send it where you want as opposed to using the script.
# > A problem with the script at present is that it does not allow to
# > directly use what's in the CVS baseline.
#
# True. I usually make a 'cvs update' in the root anyway before I go ahead and
# play. I didn't want to add that dependency in maketgz, since it builds
# tarballs, not necessarily tarballs with the very latest sources from the CVS
# baseline. It is up to you to pick 'em.
Ok. Then I'll make the CVS an optional feature and leave the original
funtionality.
# > It also does not support tagging the CVS files that make up the
# > distribution.
#
# True. But if you don't do any magic tricks, and if you're using the latest
# and fanciest sources you would simply commit everything and tag them just
# after you'd run maketgz. I've done that numerous of times.
My problem is that I want to have this as a fully automated operation
so there are no mistakes.
# > Also the script modifies patchlevel.h that is created as part of the
# > distribution. That new patchlevel.h is then not consistent with what is
# > in the CVS tree.
#
# ... if you're not committing, that is true yes. I usually commit it after
# the maketgz has run. Having the script do this modification makes it
# absolutely sure we never release a tarball with the wrong version number
# string inside.
Right.
# It may not be the perfect tool yet and it may need more tweaking, I still
# believe that having a tool like this that builds the archives is far better
# than lots of manual steps that have to be done without mistakes every time.
I totally agree. ;)
# I'm not convinced that such a strict connection between maketgz and the CVS
# management is needed. But I'm not very much against it either. ;-)
And if it works both ways.... :)
-- Kent Landfield Phone: 1-817-545-2502 Email: kent@landfield.com http://www.landfield.com/ Email: kent@nfr.net http://www.nfr.net/ Search the Usenet FAQ Archive at http://www.faqs.org/faqs/ Search the RFC/FYI/STD/BCP Archive at http://www.faqs.org/rfcs/
![]()