cvs commit: ports/dns Makefile ports/dns/bind93 Makefile
distinfo pkg-descr pkg-message pkg-plist
eikemeier at fillmore-labs.com
Wed Aug 11 18:39:37 PDT 2004
Doug Barton wrote:
> On Thu, 12 Aug 2004, Oliver Eikemeier wrote:
>> Ok, could you just outline to me why
>> does not apply, why dns/bind93 is not related to dns/bind9 and why
>> there is no history to preserve, especially given that you mean to
>> remove dns/bind9.
> Thank you, that is a reasonable question, and deserves a reasonable
> answer. There are two main reasons why a repo copy isn't needed here.
> First, when the BIND 9 port was created originally they made the same
> mistake you are suggesting, namely reop copying the BIND 8 port. This
> brought in tons of pointless history that wasn't relevant. I want to
> remedy that problem by creating a new port that doesn't drag along
> unrelated history.
The commit long from Satoshi Asami says:
Temporarily remove bind9, it is a repo-copy from bind8 not updated.
The history is safe, so just "cvs add" the files back when bind9 is
ready to be committed.
So he must thought that the history was relevant. Don't you think
preserving the history was intended by him?
> Second, there is quite frankly no significant history _of the BIND 9
> port_ to preserve. If you had bothered to check the CVS history instead
> of simply offering a knee-jerk reaction then you would have already
> known this. Alternatively, if you did check the history, and there is
> something there that you think needs to be preserved at all costs,
> please state your case so that we can discuss it.
Since the upgrade to 9.x we had 40 revisions of the ports Makefile.
Doing a diff of dns/bind9 and dns/bind93 I see so much similarities that
I believe the bind93 port is an updated version of the bind9 port. You
might consider revision history unimportant, but "the FreeBSD Project
considers this history very useful":
> There is also (I think) a valid question of what is a "related" version
> here. BIND 9 is a complete rewrite, with a completely different
> codebase, distribution scheme, etc. BIND 9.3.x is essentially another
> rewrite, although they do share a common file format for the
> distribution, etc. If the original bind9 port had been created new
> rather than repo copied then I could more easily be persuaded that the
> bind93 port should be repo copied. However, at this point in the
> process I see no reason to drag all of that useless history into a new
Most of the bind9 port history concerns bind 9.x, and the code base
doesn't matter, it is the port that does. The bind9 and bind93 ports
seem very similar to me, or did you write the bind93 port from scratch
instead of copying end editing the bind9 port?
More information about the cvs-ports