[PATCH v2] debian: package ruby bindings
2013/5/16 David Bremner :
> Felipe Contreras writes:
>>
>> Yes. Do you have an old version of notmuch installed?
>>
>
> Yes I do. Hmm. It still seems like a bug to me; shouldn't the include
> paths be ordered to find the right include files first?
>
> d
Well I must accept I haven't spent much time on thinking how to make
extconf.rb perfect.
Excerpt from extconf.rb from bindings/ruby:
find_header('notmuch.h', '../../lib')
find_library('notmuch', 'notmuch_database_create', '../../lib')
Two problems here:
1. The library must be built before the bindings are built otherwise
the installed version is used.
However the installed version should never be used (or must be
made a non-default option somehow)
2. The library must be built in the source tree. Out-of-source tree
builds will cause extconf.rb to look for the installed version too.
Not sure whether this is a major concern because I don't know
whether out-of-tree builds are supported by the build system.
Having said that, I will try to have a look at solving these issues
this weekend.
Unfortunately I can't make promises because real-life is killing all
my time lately.
-alip
[PATCH] build: fix out-of-tree builds
On Thu, May 16 2013, David Bremner wrote: > Jani Nikula writes: > >> On Wed, 15 May 2013, David Bremner wrote: >>> Jani Nikula writes: # List all subdirectories here. Each contains its own Makefile.local -subdirs := compat completion emacs lib man parse-time-string -subdirs += performance-test util test +subdirs = compat completion emacs lib man parse-time-string performance-test util test >>> >>> Is this a functional change or a cosmetic one? >> >> Functional. All of it seems to be required for out-of-tree builds (*) to >> work. I'm afraid I've forgotten exactly why both := and += make it fail >> (I wrote the patch a few months ago). >> > > Still I guess it deserves a comment so we don't break it again? The following options could be tested (some day in (distant?) future): subdirs = compat completion emacs lib man parse-time-string subdirs += performance-test util test or SUBDIRS := compat completion emacs lib man parse-time-string SUBDIRS += performance-test util test subdirs = $(SUBDIRS) but as Jani has already tested his version works just adding that comment David asked is the least labor-intensive thing to do -- and therefore IMHO the best thing to do for this case (for sure I'm not going to spend more time on this now :D). > d Tomi
[PATCH v2] debian: package ruby bindings
Felipe Contreras writes: > > Yes. Do you have an old version of notmuch installed? > Yes I do. Hmm. It still seems like a bug to me; shouldn't the include paths be ordered to find the right include files first? d
[PATCH v2] debian: package ruby bindings
Felipe Contreras writes: > Signed-off-by: Felipe Contreras > --- > debian/control | 14 ++ > debian/notmuch-ruby.install | 1 + > debian/rules| 3 +++ > 3 files changed, 18 insertions(+) > create mode 100644 debian/notmuch-ruby.install > I get a build failure here; I installed ruby-dev and ran the usual "make debian-snapshot". Does it work for you in a clean with only the declared build-depends installed? cd bindings/ruby && ruby extconf.rb --vendor && make checking for notmuch.h in ../../lib... yes checking for notmuch_database_create() in -lnotmuch... yes creating Makefile make[2]: Entering directory `/home/bremner/software/upstream/notmuch/bindings/ruby' compiling status.c status.c: In function ?notmuch_rb_status_raise?: status.c:46:10: error: ?NOTMUCH_STATUS_UNBALANCED_ATOMIC? undeclared (first use in this function) status.c:46:10: note: each undeclared identifier is reported only once for each function it appears in make[2]: *** [status.o] Error 1 make[2]: Leaving directory `/home/bremner/software/upstream/notmuch/bindings/ruby' make[1]: *** [override_dh_auto_build] Error 2 make[1]: Leaving directory `/home/bremner/software/upstream/notmuch' make: *** [build] Error 2 dpkg-buildpackage: error: debian/rules build gave error exit status 2 debuild: fatal error at line 1357: dpkg-buildpackage -rfakeroot -D -us -uc -i failed
[PATCH v2] debian: package ruby bindings
On Thu, May 16, 2013 at 9:31 AM, David Bremner wrote: > Felipe Contreras writes: >> >> Yes. Do you have an old version of notmuch installed? >> > > Yes I do. Hmm. It still seems like a bug to me; shouldn't the include > paths be ordered to find the right include files first? Yes, it is a bug. And there were attempts to fix it, but it broke the normal build AFAICR. -- Felipe Contreras
[PATCH v2] debian: package ruby bindings
On Thu, May 16, 2013 at 8:51 AM, David Bremner wrote: > Felipe Contreras writes: > >> Signed-off-by: Felipe Contreras >> --- >> debian/control | 14 ++ >> debian/notmuch-ruby.install | 1 + >> debian/rules| 3 +++ >> 3 files changed, 18 insertions(+) >> create mode 100644 debian/notmuch-ruby.install >> > > I get a build failure here; I installed ruby-dev and ran the usual > "make debian-snapshot". > > Does it work for you in a clean with only the declared build-depends > installed? Yes. Do you have an old version of notmuch installed? -- Felipe Contreras
[PATCH] build: fix out-of-tree builds
Jani Nikula writes: > On Wed, 15 May 2013, David Bremner wrote: >> Jani Nikula writes: >>> # List all subdirectories here. Each contains its own Makefile.local >>> -subdirs := compat completion emacs lib man parse-time-string >>> -subdirs += performance-test util test >>> +subdirs = compat completion emacs lib man parse-time-string >>> performance-test util test >> >> Is this a functional change or a cosmetic one? > > Functional. All of it seems to be required for out-of-tree builds (*) to > work. I'm afraid I've forgotten exactly why both := and += make it fail > (I wrote the patch a few months ago). > Still I guess it deserves a comment so we don't break it again? d
Re: [PATCH v2] debian: package ruby bindings
On Thu, May 16, 2013 at 9:31 AM, David Bremner wrote: > Felipe Contreras writes: >> >> Yes. Do you have an old version of notmuch installed? >> > > Yes I do. Hmm. It still seems like a bug to me; shouldn't the include > paths be ordered to find the right include files first? Yes, it is a bug. And there were attempts to fix it, but it broke the normal build AFAICR. -- Felipe Contreras ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: [PATCH v2] debian: package ruby bindings
2013/5/16 David Bremner :
> Felipe Contreras writes:
>>
>> Yes. Do you have an old version of notmuch installed?
>>
>
> Yes I do. Hmm. It still seems like a bug to me; shouldn't the include
> paths be ordered to find the right include files first?
>
> d
Well I must accept I haven't spent much time on thinking how to make
extconf.rb perfect.
Excerpt from extconf.rb from bindings/ruby:
find_header('notmuch.h', '../../lib')
find_library('notmuch', 'notmuch_database_create', '../../lib')
Two problems here:
1. The library must be built before the bindings are built otherwise
the installed version is used.
However the installed version should never be used (or must be
made a non-default option somehow)
2. The library must be built in the source tree. Out-of-source tree
builds will cause extconf.rb to look for the installed version too.
Not sure whether this is a major concern because I don't know
whether out-of-tree builds are supported by the build system.
Having said that, I will try to have a look at solving these issues
this weekend.
Unfortunately I can't make promises because real-life is killing all
my time lately.
-alip
___
notmuch mailing list
notmuch@notmuchmail.org
http://notmuchmail.org/mailman/listinfo/notmuch
Re: [PATCH v2] debian: package ruby bindings
Felipe Contreras writes: > > Yes. Do you have an old version of notmuch installed? > Yes I do. Hmm. It still seems like a bug to me; shouldn't the include paths be ordered to find the right include files first? d ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: [PATCH v2] debian: package ruby bindings
On Thu, May 16, 2013 at 8:51 AM, David Bremner wrote: > Felipe Contreras writes: > >> Signed-off-by: Felipe Contreras >> --- >> debian/control | 14 ++ >> debian/notmuch-ruby.install | 1 + >> debian/rules| 3 +++ >> 3 files changed, 18 insertions(+) >> create mode 100644 debian/notmuch-ruby.install >> > > I get a build failure here; I installed ruby-dev and ran the usual > "make debian-snapshot". > > Does it work for you in a clean with only the declared build-depends > installed? Yes. Do you have an old version of notmuch installed? -- Felipe Contreras ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: [PATCH v2] debian: package ruby bindings
Felipe Contreras writes: > Signed-off-by: Felipe Contreras > --- > debian/control | 14 ++ > debian/notmuch-ruby.install | 1 + > debian/rules| 3 +++ > 3 files changed, 18 insertions(+) > create mode 100644 debian/notmuch-ruby.install > I get a build failure here; I installed ruby-dev and ran the usual "make debian-snapshot". Does it work for you in a clean with only the declared build-depends installed? cd bindings/ruby && ruby extconf.rb --vendor && make checking for notmuch.h in ../../lib... yes checking for notmuch_database_create() in -lnotmuch... yes creating Makefile make[2]: Entering directory `/home/bremner/software/upstream/notmuch/bindings/ruby' compiling status.c status.c: In function ‘notmuch_rb_status_raise’: status.c:46:10: error: ‘NOTMUCH_STATUS_UNBALANCED_ATOMIC’ undeclared (first use in this function) status.c:46:10: note: each undeclared identifier is reported only once for each function it appears in make[2]: *** [status.o] Error 1 make[2]: Leaving directory `/home/bremner/software/upstream/notmuch/bindings/ruby' make[1]: *** [override_dh_auto_build] Error 2 make[1]: Leaving directory `/home/bremner/software/upstream/notmuch' make: *** [build] Error 2 dpkg-buildpackage: error: debian/rules build gave error exit status 2 debuild: fatal error at line 1357: dpkg-buildpackage -rfakeroot -D -us -uc -i failed ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: [PATCH] build: fix out-of-tree builds
On Thu, May 16 2013, David Bremner wrote: > Jani Nikula writes: > >> On Wed, 15 May 2013, David Bremner wrote: >>> Jani Nikula writes: # List all subdirectories here. Each contains its own Makefile.local -subdirs := compat completion emacs lib man parse-time-string -subdirs += performance-test util test +subdirs = compat completion emacs lib man parse-time-string performance-test util test >>> >>> Is this a functional change or a cosmetic one? >> >> Functional. All of it seems to be required for out-of-tree builds (*) to >> work. I'm afraid I've forgotten exactly why both := and += make it fail >> (I wrote the patch a few months ago). >> > > Still I guess it deserves a comment so we don't break it again? The following options could be tested (some day in (distant?) future): subdirs = compat completion emacs lib man parse-time-string subdirs += performance-test util test or SUBDIRS := compat completion emacs lib man parse-time-string SUBDIRS += performance-test util test subdirs = $(SUBDIRS) but as Jani has already tested his version works just adding that comment David asked is the least labor-intensive thing to do -- and therefore IMHO the best thing to do for this case (for sure I'm not going to spend more time on this now :D). > d Tomi ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: [PATCH] build: fix out-of-tree builds
Jani Nikula writes: > On Wed, 15 May 2013, David Bremner wrote: >> Jani Nikula writes: >>> # List all subdirectories here. Each contains its own Makefile.local >>> -subdirs := compat completion emacs lib man parse-time-string >>> -subdirs += performance-test util test >>> +subdirs = compat completion emacs lib man parse-time-string >>> performance-test util test >> >> Is this a functional change or a cosmetic one? > > Functional. All of it seems to be required for out-of-tree builds (*) to > work. I'm afraid I've forgotten exactly why both := and += make it fail > (I wrote the patch a few months ago). > Still I guess it deserves a comment so we don't break it again? d ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
