From: Doug Barton on 18 Jul 2010 22:12 On 07/18/10 15:38, Mark Linimon wrote: > The intention of bsd.perl.mk was to eventually allow a bunch of code > to be pulled out of bsd.port.mk, and to have bsd.perl.mk only included > conditionally, on the theory that it will speed up INDEX building > somewhat. (I have not tested for speedup). Sounds like something that should be tested. > The problem is that there are N ports that assume that the logic in > bsd.perl.mk is always available. I've tried to convince people that > these are bugs, but OTOH if you leave out one of these definitions > such as USE_PERL5 or PERL_CONFIGURE, and _PERL_REFACTORING_COMPLETE > is defined, then INDEX breaks. I'm confused. Are you saying that there are ports that try to use stuff that's defined in bsd.perl.mk without including it? If so, that should be fixed. Also, how many ports are we talking about here? What is duplicated between .mk files? Is there an easy algorithm to determine this? > Every once in a while I try to put in patches to force N to zero, but > then I get stuck on wierd edge cases. This is why the switch has not > been thrown. Step 1: Fix the easy cases Step 2: Notify maintainers of wacky edge cases (perhaps followed by a brief delay to let them fix it themselves) Step 2: Disconnect broken wacky edge cases from the build Step 3: Reconnect wacky edge cases as they get fixed At bare minimum Step 1 should be followed immediately to avoid people copy/pasting bad examples. > Every time I try to work on this, "something happens" and it gets > shoved on the back-burner for several months. This has been going > on for several years now ... Then it's way beyond time you asked for help. :) I've cc'ed perl@ in case they are interested in this. If not, I might be. > At this point it may be better to just do the following: > > - unconditionally include bsd.perl.mk and get rid of the code > duplication that is in bsd.port.mk. Um, no. That's 100% backwards. If you're going to include it unconditionally there is no point in having a separate file. But I don't think that including it unconditionally is the right answer, it would be better to fix this properly. Doug -- Improve the effectiveness of your Internet presence with a domain name makeover! http://SupersetSolutions.com/ Computers are useless. They can only give you answers. -- Pablo Picasso _______________________________________________ freebsd-ports(a)freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscribe(a)freebsd.org"
From: "Philip M. Gollucci" on 19 Jul 2010 18:22 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I've long been interested in this. I must be using the wrong options or something. About 1 yr ago I was able to build the index after removing the appropriate things in Mk/. On 07/19/10 02:12, Doug Barton wrote: > On 07/18/10 15:38, Mark Linimon wrote: >> The intention of bsd.perl.mk was to eventually allow a bunch of code >> to be pulled out of bsd.port.mk, and to have bsd.perl.mk only included >> conditionally, on the theory that it will speed up INDEX building >> somewhat. (I have not tested for speedup). > > Sounds like something that should be tested. > >> The problem is that there are N ports that assume that the logic in >> bsd.perl.mk is always available. I've tried to convince people that >> these are bugs, but OTOH if you leave out one of these definitions >> such as USE_PERL5 or PERL_CONFIGURE, and _PERL_REFACTORING_COMPLETE >> is defined, then INDEX breaks. > > I'm confused. Are you saying that there are ports that try to use stuff > that's defined in bsd.perl.mk without including it? If so, that should > be fixed. > > Also, how many ports are we talking about here? What is duplicated > between .mk files? Is there an easy algorithm to determine this? > >> Every once in a while I try to put in patches to force N to zero, but >> then I get stuck on wierd edge cases. This is why the switch has not >> been thrown. > > Step 1: Fix the easy cases > Step 2: Notify maintainers of wacky edge cases (perhaps followed by a > brief delay to let them fix it themselves) > Step 2: Disconnect broken wacky edge cases from the build > Step 3: Reconnect wacky edge cases as they get fixed > > At bare minimum Step 1 should be followed immediately to avoid people > copy/pasting bad examples. > >> Every time I try to work on this, "something happens" and it gets >> shoved on the back-burner for several months. This has been going >> on for several years now ... > > Then it's way beyond time you asked for help. :) I've cc'ed perl@ in > case they are interested in this. If not, I might be. > >> At this point it may be better to just do the following: >> >> - unconditionally include bsd.perl.mk and get rid of the code >> duplication that is in bsd.port.mk. > > Um, no. That's 100% backwards. If you're going to include it > unconditionally there is no point in having a separate file. But I don't > think that including it unconditionally is the right answer, it would be > better to fix this properly. > > > Doug > - -- - ------------------------------------------------------------------------ 1024D/DB9B8C1C B90B FBC3 A3A1 C71A 8E70 3F8C 75B8 8FFB DB9B 8C1C Philip M. Gollucci (pgollucci(a)p6m7g8.com) c: 703.336.9354 VP Apache Infrastructure; Member, Apache Software Foundation Committer, FreeBSD Foundation Consultant, P6M7G8 Inc. Sr. System Admin, Ridecharge Inc. Work like you don't need the money, love like you'll never get hurt, and dance like nobody's watching. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.14 (FreeBSD) iD8DBQFMRNAcdbiP+9ubjBwRAtS3AJ9rCBCA3u3N7QEoiov5qfHTBXVVCQCgjtEr j47qDeZUTQrqFZpVCdiFXEU= =kroi -----END PGP SIGNATURE----- _______________________________________________ freebsd-ports(a)freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscribe(a)freebsd.org"
From: "Philip M. Gollucci" on 19 Jul 2010 23:59 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 07/20/10 03:57, Doug Barton wrote: > On Mon, 19 Jul 2010, Philip M. Gollucci wrote: > >> I've long been interested in this. I must be using the wrong options or >> something. About 1 yr ago I was able to build the index after removing >> the appropriate things in Mk/. > > Well, seems like I have good news. :) I ran tests on both 9-current and > 6-stable with a full, clean tree and no make.conf; with and without > -D_PERL_REFACTORING_COMPLETE and the INDEX files were identical. So it > looks like this is a good time to pull the trigger and make the change. > > If portmgr approves I'll be glad to make the change myself, but I > imagine that a pointyhat run first is in order. The diff is easy, but > I've attached it just in case. > > > Doug > You didn't quite test everything. Some ports have optional PERL dependencies. You should probably try WITH_PERL=yes before and after too. - -- - ------------------------------------------------------------------------ 1024D/DB9B8C1C B90B FBC3 A3A1 C71A 8E70 3F8C 75B8 8FFB DB9B 8C1C Philip M. Gollucci (pgollucci(a)p6m7g8.com) c: 703.336.9354 VP Apache Infrastructure; Member, Apache Software Foundation Committer, FreeBSD Foundation Consultant, P6M7G8 Inc. Sr. System Admin, Ridecharge Inc. Work like you don't need the money, love like you'll never get hurt, and dance like nobody's watching. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.14 (FreeBSD) iD8DBQFMRR8GdbiP+9ubjBwRAjiKAKCJ3obJGzmJu/xnadw0IVck08U+VACgijF1 eQFksq8c7TASyK+SA7WnvAY= =Tko5 -----END PGP SIGNATURE----- _______________________________________________ freebsd-ports(a)freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscribe(a)freebsd.org"
From: "Philip M. Gollucci" on 20 Jul 2010 01:28 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 07/20/10 04:52, Doug Barton wrote: > So yeah, a pointyhat run is almost certainly a good idea before actually > making the change, however 'make index' was the canary in Mark's > original coal mine, and fortunately that bird is still singing. :) Excellent, then your findings and mine agree. So unless we've missed something would do the honors to send a pr in for an -exp run. CC me please. - -- - ------------------------------------------------------------------------ 1024D/DB9B8C1C B90B FBC3 A3A1 C71A 8E70 3F8C 75B8 8FFB DB9B 8C1C Philip M. Gollucci (pgollucci(a)p6m7g8.com) c: 703.336.9354 VP Apache Infrastructure; Member, Apache Software Foundation Committer, FreeBSD Foundation Consultant, P6M7G8 Inc. Sr. System Admin, Ridecharge Inc. Work like you don't need the money, love like you'll never get hurt, and dance like nobody's watching. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.14 (FreeBSD) iD8DBQFMRTQJdbiP+9ubjBwRAtTKAJ4h8eWOgY4z+85gXYv+YUHV3Epx3QCeI1VA vXpiWOlWwBVw8p3eTdHfC0Q= =HvWi -----END PGP SIGNATURE----- _______________________________________________ freebsd-ports(a)freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscribe(a)freebsd.org"
From: Doug Barton on 20 Jul 2010 00:52
On Tue, 20 Jul 2010, Philip M. Gollucci wrote: > You didn't quite test everything. I didn't try. :) In talking to Mark his previous testing clearly demonstrated that with the knob set would cause the INDEX generation to fail spectacularly. So the fact that this did not happen, seems like a good sign. > Some ports have optional PERL dependencies. > > You should probably try WITH_PERL=yes before and after too. That's a great suggestion, thanks. I re-ran the tests, all 4 with WITH_PERL defined, and then before and after _PERL_REFACTORING_COMPLETE and the 2 WITH_PERL INDEX files were the same in each branch. Different than the ones generated without WITH_PERL obviously ... So yeah, a pointyhat run is almost certainly a good idea before actually making the change, however 'make index' was the canary in Mark's original coal mine, and fortunately that bird is still singing. :) Doug -- Improve the effectiveness of your Internet presence with a domain name makeover! http://SupersetSolutions.com/ Computers are useless. They can only give you answers. -- Pablo Picasso _______________________________________________ freebsd-ports(a)freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscribe(a)freebsd.org" |