FreeBSD Bugzilla – Bug 165767
[maintainer patch] astro/merkaartor update to 0.18-dev in order to fix build with boost-1.49
Last modified: 2012-03-16 21:20:00 UTC
From this pr: http://www.freebsd.org/cgi/query-pr.cgi?pr=ports/156253 (an update for boost) an e-mail found its way to me warning me of breakage and requesting action.
This update updates to the latest Merkaartor snapshot from git. This means that Merkaartor no longer depends on boost. Sadly, as upstream development is not very active at this point in time, there is no real release for 0.18 now or in the foreseeable future. Despite being a snapshot, the program is stable.
The snapshot tarball is hosted by me.
The alternative to build 0.17.2 with built-in boost (1.38) does not work reliably with a newer boost installed on the system.
Scratch that patch, a better one is underway
- tweak PORTVERSION to prevent confusion with future releases,
- remove proj and gdal from options as they are now required.
I'll take it
Here some feedback. First, my tinderbox swallowed your patch very well and =
everything seems to compile perfectly. Regarding the new snapshot, I have t=
he following one suggestion:
It seems to me that we should bump PORTEPOCH and then use a more visible po=
rt version, e.g. date (20120315) or first digits of git revision, etc.
What do you think?
Op zo 11 mrt 2012 21:06:44 schreef Carlo Strub:
> Here some feedback. First, my tinderbox swallowed your patch very well and
> everything seems to compile perfectly. Regarding the new snapshot, I have
> the following one suggestion:
> It seems to me that we should bump PORTEPOCH and then use a more visible
> port version, e.g. date (20120315) or first digits of git revision, etc.
> What do you think?
> Carlo Strub
> Ports committer
From what I've read PORTEPOCH should not be set unless absolutely necessary (to
fix a mistake or something:
So I'd rather not set that.
I took the example here:
for marking a port alpha, but I guess 0.18 < 0.18.a so when a real release
happens you couldn't call it 0.18, but would have to call it 0.18.b or 0.18.c,
I'd be happy to call it 0.17.20120304 (the checkout date), but the program runs
as 0.18, I could edit Config.pri further to correct that, but I fear that would
only cause further confusement, especially when a user asks questions upstream.
Another option then is to call the port 0.18.20120304 and for any subsequent
release in the 0.18 branch maintain and update the date suffix. The one patch
the port now has, adds "-dev" to the windowtitle and the about-dialog to remind
the user that (s)he is using a development version, that patch can be dropped
when a real release is made.
Thinking about it, I favor the last option. If you agree, I'll adjust the patch
and the distfile accordingly.
Hmm. This is a tricky question to answer correctly. On the one hand you are=
right that PORTEPOCH should not be used extensively, on the other hand nam=
ing it 18.a may be misleading users in thinking that this was an official a=
lpha release. To be honest, I have no super strong opinion on it. Let me co=
ntact my mentors to ask for advise. I'll get back to you. In case we have d=
ecided on a good solution you don't need to resend the patch. I can fix tha=
Running some tests with pkg_version -vL= and PORTVERSION, the ports versioning
system actually handles those letters as you would expect from the examples.
merkaartor-0.18.a < needs updating (port has 0.18.20120304)
merkaartor-0.18.a < needs updating (port has 0.18)
merkaartor-0.18.20120304 > succeeds port (port has 0.18)
merkaartor-0.18.a.20120304 < needs updating (port has 0.18)
So using 0.18.a would work when a 0.18 is released later,
using 0.18.20120304 would not work unless you maintain and update the date
suffix until 0.19 arrives.
0.18.a.20120304 would also work and I think I like it best. The 'a' tells the
ports system that this stuff is pre-release, adding a date string is a sign to
the user that something "special" has been done. Once a 0.18 is released,
simply naming it that will obsolete any and all 0.18.a versions.
-----BEGIN PGP SIGNED MESSAGE-----
As Carlo's mentor, he asked me to review this PR for my opinion on the
version string convention. From the looks of it, you've both narrowed
it down to a version string that's constructed similarly to other ones
already in the ports tree. I ran a quick analysis:
# cut -f1 -d\| /usr/ports/INDEX-8 | egrep "\.[ab]\.?[[:digit:]]+" | sort
| grep 201
There is a handful of ports that include a ".a" or ".b" as well as a
date stamp. If you invert the grep, you'll find there are 100+ that
only use ".a" or ".b" with a simple incrementing number.
Each form is acceptable, with one more commonly used than the other.
http://www.FreeBSD.org/ - The Power To Serve
http://www.sourcehosting.net/ - Ready. Set. Code.
http://twitter.com/cpucycle/ - Follow you, follow me
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
-----END PGP SIGNATURE-----
cs 2012-03-16 21:12:17 UTC
FreeBSD ports repository
astro/merkaartor Makefile distinfo pkg-plist
- Update to last git snapshot version (no official release for a long time)
- Remove boost dependency
- Remove GDAL and PROJ options as they are now required
Submitted by: Oliver Heesakkers <firstname.lastname@example.org> (maintainer)
Approved by: glarkin@ (mentor, implicit)
Feature safe: yes
Revision Changes Path
1.21 +5 -20 ports/astro/merkaartor/Makefile
1.9 +2 -2 ports/astro/merkaartor/distinfo
1.1 +10 -0 ports/astro/merkaartor/files/patch-src__Config.pri (new)
1.2 +0 -11 ports/astro/merkaartor/files/patch-src__Sync__DownloadOSM.cpp (dead)
1.6 +2 -3 ports/astro/merkaartor/pkg-plist
email@example.com mailing list
To unsubscribe, send any mail to "firstname.lastname@example.org"
Committed. Thank you very much.