- Fix PKGORIGIN Generated with FreeBSD Port Tools 0.99_6 (mode: change, diff: ports)
Responsible Changed From-To: freebsd-ports-bugs->rm I will take it.
Hi koobs, In first may I ask you to not add py27 prefix when filling pr next time? :). GNATS tries (in simple-case) to auto-assign pr by checking this line for portname. There is no such port security/py27-oauthlib, but there is security/py-oauthlib. Please just omit this 27 prefix next time. Thank you in advance. I just reviewed the port and want to suggest some changes. Here is the patch, that: - adds some tabs in Makefile header for aligning - removes LICENSE_FILE knob, because it doesn't needed for well-known licenses - py-rsa needs both at build time (for tests, if they are used) and run time (for signature module), so it adds py-rsa to RUN_DEPENDS too. - marks this port python3 ready The last and the most contradictory - why you need to unpack the egg before installing it? In my patch I removed all the custom easy_install args and moved pkg-plist into a Makefile:PLIST_FILES. It works just fine, but I'm not sure, may be you had a specific reason to do so.. What do you say? Do you approve? -- Regards, Ruslan Tinderboxing kills... the drives.
rm 2012-06-14 14:28:25 UTC FreeBSD ports repository Modified files: security/py-oauthlib Makefile Log: - correct port category PR: 169068 Submitted by: koobs <koobs.freebsd at gmail dot com> (maintainer) Revision Changes Path 1.2 +1 -1 ports/security/py-oauthlib/Makefile _______________________________________________ cvs-all@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/cvs-all To unsubscribe, send any mail to "cvs-all-unsubscribe@freebsd.org"
State Changed From-To: open->feedback Original patch committed. Waiting for maintainer's feedback about proposed changes.
Absolutely approve, I'm the padawan here :) To address each point individually: > In first may I ask you to not add py27 prefix when filling pr next time? = :). I use porttools `port submit` to send-pr's and it is generating the Synopsis field. In this case I forgot to remove '27'. Perhaps a patch to omit *_PKGNAMEPREFIX from the generated Synopsis would be handy here? > I just reviewed the port and want to suggest some changes. Here is the > patch, that: > - adds some tabs in Makefile header for aligning Again i'm using the standard porttools port create template here. I assumed defaults were good, clean and safe :) > - removes LICENSE_FILE knob, because it doesn't needed for well-known > =C2=A0licenses Learned this last week and not doing this for BSD/GPLvX/MIT anymore. What constitutes 'well-known' ? > - py-rsa needs both at build time (for tests, if they are used) and run > =C2=A0time (for signature module), so it adds py-rsa to RUN_DEPENDS too. > - marks this port python3 ready Good to know, thank you, although I'm not 100% on this. What if the tests require non-stldlib nose/someotherpymod, do I BUILD_DEPEND these too? Or PORT_OPTIONS MTESTS to conditionally add it? > The last and the most contradictory - why you need to unpack the egg befo= re > installing it? In my patch I removed all the custom easy_install args and > moved pkg-plist into a Makefile:PLIST_FILES. It works just fine, but I'm = not > sure, may be you had a specific reason to do so.. There is a bigger conversation happening about adding default -Z args in bsd.python.mk and moving onto py-distribute among other things. Alot of python modules do not set zip_safe=3DFalse when they should and there's no easy way to preffix-override PYEASINSTALL_ARGS. That leaves the choice of this, or patch setup.py. Having thought about it, i'll go the patch setup.py route from here on unless you have ideas? > What do you say? Do you approve? Absolutely, and thank you for the feedback Regards, Koobs
koobs k wrote on 15.06.2012 12:55: > I use porttools `port submit` to send-pr's and it is generating the > Synopsis field. In this case I forgot to remove '27'. Perhaps a patch > to omit *_PKGNAMEPREFIX from the generated Synopsis would be handy > here? Understood. I didn't know about this porttools feature and believe this is incorrect. I'll look at this and tabs issue and will submit it as an pr, thanks for clarifying :). >> - removes LICENSE_FILE knob, because it doesn't needed for well-known >> licenses > > Learned this last week and not doing this for BSD/GPLvX/MIT anymore. > What constitutes 'well-known' ? You may treat it as `popular' I believe, this is about my poor English skills. Generally, it's safe to omit this knob for any license listed in Mk/bsd.licenses.db.mk > >> - py-rsa needs both at build time (for tests, if they are used) and run >> time (for signature module), so it adds py-rsa to RUN_DEPENDS too. >> - marks this port python3 ready > > Good to know, thank you, although I'm not 100% on this. What if the > tests require non-stldlib nose/someotherpymod, do I BUILD_DEPEND these > too? Or PORT_OPTIONS MTESTS to conditionally add it? I just grep'd for rsa and see that is used signature.py (hence runtime dependency) and for build it is only needed in tests. It's safe to remove it from BUILD_DEPENDS at all, because we have not (yet) any automatically tests running for python ports. If port testing framework using nose, than tests will fail if it's not installed on build stage. But, as I wright before, it's not important right now, because packaging will succeed anyway. But there is one caveat - if the port has support for building documentation (with py-sphinx for example), then someone need to check port build with py-sphinx installed, because packing list may (and in 80% cases will) changes because of the docs installed. > >> The last and the most contradictory - why you need to unpack the egg before >> installing it? In my patch I removed all the custom easy_install args and >> moved pkg-plist into a Makefile:PLIST_FILES. It works just fine, but I'm not >> sure, may be you had a specific reason to do so.. > > There is a bigger conversation happening about adding default -Z args > in bsd.python.mk and moving onto py-distribute among other things. > Alot of python modules do not set zip_safe=False when they should and > there's no easy way to preffix-override PYEASINSTALL_ARGS. To be short, I dislike the idea and see no real reason to change something. I have no problems with unpacked .egg's and believe that any problems should be resolved in individual manner. > That leaves the choice of this, or patch setup.py. Having thought > about it, i'll go the patch setup.py route from here on unless you > have ideas? I will revert this part of my patch and commit just what is left :). Let's see if -Z stuff will be committed into the tree, and they operate accordingly of that. -- Regards, Ruslan Tinderboxing kills... the drives.
rm 2012-06-15 11:22:44 UTC FreeBSD ports repository Modified files: security/py-oauthlib Makefile pkg-plist Log: - remove unneded LICENSE_FILE knob - add dependency on py-rsa to RUN_DEPENDS too - mark this port python3 ready - remove trailing slash from pkg-plist entry - bump PORTREVISION because of dependency change PR: 169068 (follow-up patch) Submitted by: rm (myself) Approved by: koobs k <koobs.freebsd at gmail dot com> (maintainer) Revision Changes Path 1.3 +6 -4 ports/security/py-oauthlib/Makefile 1.2 +2 -2 ports/security/py-oauthlib/pkg-plist _______________________________________________ cvs-all@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/cvs-all To unsubscribe, send any mail to "cvs-all-unsubscribe@freebsd.org"
State Changed From-To: feedback->closed Committed, thank you!