perlgssapi-developer Mailing List for Perl GSSAPI bindings (Page 7)
Brought to you by:
achimgrolms
Menu
▾
▴
perlgssapi-developer — discussion of development process
You can subscribe to this list here.
| 2006 |
Jan
|
Feb
(99) |
Mar
(17) |
Apr
(8) |
May
(6) |
Jun
|
Jul
(1) |
Aug
(2) |
Sep
|
Oct
|
Nov
|
Dec
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2007 |
Jan
|
Feb
(2) |
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
| 2008 |
Jan
(10) |
Feb
(6) |
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
(1) |
Sep
|
Oct
|
Nov
|
Dec
|
| 2009 |
Jan
|
Feb
(1) |
Mar
|
Apr
|
May
(4) |
Jun
|
Jul
|
Aug
(4) |
Sep
|
Oct
|
Nov
|
Dec
|
| 2012 |
Jan
(1) |
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
| 2013 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
(2) |
Oct
|
Nov
|
Dec
|
| 2014 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
(1) |
Dec
|
|
From: Merijn B. <me...@il...> - 2006-02-08 12:32:15
|
Quoting Leif Johansson (le...@it...):
>
> Right - I was referring to the symbolic links you mentioned to get
> around having to rebuild dependencies on (in this case) kerberos
> when you rebuild kerberos.
>
Oh yeah ofcourse, the ABI can be incompatible and actually is between
MIT Kerberos 1.2, 1.3 and 1.4. But minor revision numbers so far have
not been an issue. I make the Perl module available in multiple versions
as well, and incorporate the MIT release in the overal version. So it
will be GSSAPI 0.15-1.4 for example. Too much detail :-)
Cheers,
--
Merijn Broeren | Sometime in the middle ages, God got fed up with us
Software Geek | and put earth at sol.milky-way.univ in his kill-file.
| Pray all you want, it just gets junked.
|
|
From: Leif J. <le...@it...> - 2006-02-08 12:19:00
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Merijn Broeren wrote: > Quoting Leif Johansson (le...@it...): >> I think it is a good ide although as a person who also does his own >> software packaging I should warn you that second-guessing the linker >> that way is a loaded gun pointed at your foot. But you already know >> that :-) >> > Well, I'm not trying to second guess the linker, just exactly doing what > krb5-config says. Sadly MakeMaker ignores the -W flags, so you have to > put them in LDDLFLAGS. It's easier if everything comes out of /usr/lib > or /usr/local/lib but in my specific case we don't use that. > > Cheers, Right - I was referring to the symbolic links you mentioned to get around having to rebuild dependencies on (in this case) kerberos when you rebuild kerberos. MVH leifj -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFD6eGq8Jx8FtbMZncRAgnjAJ98tRZUe6V+ltd0ZFv0qNyI+c8z/QCgxvgq QRnXBu56tQ60+pUegMRgYuc= =1KtE -----END PGP SIGNATURE----- |
|
From: Merijn B. <me...@il...> - 2006-02-08 12:12:25
|
Quoting Leif Johansson (le...@it...):
>
> I think it is a good ide although as a person who also does his own
> software packaging I should warn you that second-guessing the linker
> that way is a loaded gun pointed at your foot. But you already know
> that :-)
>
Well, I'm not trying to second guess the linker, just exactly doing what
krb5-config says. Sadly MakeMaker ignores the -W flags, so you have to
put them in LDDLFLAGS. It's easier if everything comes out of /usr/lib
or /usr/local/lib but in my specific case we don't use that.
Cheers,
--
Merijn Broeren | Sometime in the middle ages, God got fed up with us
Software Geek | and put earth at sol.milky-way.univ in his kill-file.
| Pray all you want, it just gets junked.
|
|
From: Leif J. <le...@it...> - 2006-02-08 12:03:28
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Merijn Broeren wrote: > Quoting Achim Grolms (per...@gr...): >> I have uploaded 0.14. >> >> I think that this is a first "make footprint" release, >> and more releases have to follow >> > Great, thanks again for picking this up. > > I've got an issue with the build process now. I like very much that I > can pass an argument to Makefile.PL which is used to find the kerberos > libraries. But for me this is right now not working, for two reasons. > One to do with MakeMaker and one with our subtle setup. > > Here is the output of Makefile.PL right now: > > Searching krb5-config > command...//ms/dist/kerberos/PROJ/mitkrb5/1.4-lib-prod/bin/krb5-config > using GSSAPI implementation Kerberos 5 release 1.4.3 > > Using libconfiguration > -L/ms/dist/kerberos/PROJ/mitkrb5/1.4.3-ms0/.exec/ia32.linux.2.4.lib6/lib > -Wl,-rpath > -Wl,/ms/dist/kerberos/PROJ/mitkrb5/1.4.3-ms0/.exec/ia32.linux.2.4.lib6/lib > -lgssapi_krb5 -lkrb5 -lk5crypto -lkrb5support -lcom_err -lresolv > > Setting MIT defines > > using own defines > > Checking if your kit is complete... > Looks good > Unrecognized argument in LIBS ignored: '-Wl,-rpath' > Unrecognized argument in LIBS ignored: '-Wl,/ms/dist/kerberos/PROJ/mitkrb5/1.4.3-ms0/.exec/ia32.linux.2.4.lib6/lib' > Writing Makefile for GSSAPI > > Which seems quite reasonable. Except that I want to use 1.4-lib-prod, > which is a symlink to the actual version 1.4.3-ms0. This symlink can be > changed at a later date, so I don't have to recompile the perl GSSAPI > module if I want to address a bug for instance. The krb5-config is too > specific (I could change that ofcourse). This is debatable and I can > agree it is just my problem to deal with. > > The second problem is more of an issue, the ignoring if the -Wl > arguments. These are needed at link time, otherwise I don't get a proper > runpath. I can set these using LDDLFLAGS as well as LIBS in > WriteMakefile. > > So I propose a mechanism that allows me (and probably other control > freaks) to set the LIBS, INC and LDDLFLAGS directly. See attached patch. > Any objections? I think it is a good ide although as a person who also does his own software packaging I should warn you that second-guessing the linker that way is a loaded gun pointed at your foot. But you already know that :-) I say go ahead! MVH leifj -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFD6d318Jx8FtbMZncRAreOAJ46udSOCRHYAOHIdskheRV25rl6PgCdFSfp K9gXppen/m79LhIGZJ5BP28= =fDd+ -----END PGP SIGNATURE----- |
|
From: Merijn B. <me...@il...> - 2006-02-08 11:37:18
|
Quoting Achim Grolms (per...@gr...):
> I have uploaded 0.14.
>
> I think that this is a first "make footprint" release,
> and more releases have to follow
>
Great, thanks again for picking this up.
I've got an issue with the build process now. I like very much that I
can pass an argument to Makefile.PL which is used to find the kerberos
libraries. But for me this is right now not working, for two reasons.
One to do with MakeMaker and one with our subtle setup.
Here is the output of Makefile.PL right now:
Searching krb5-config
command...//ms/dist/kerberos/PROJ/mitkrb5/1.4-lib-prod/bin/krb5-config
using GSSAPI implementation Kerberos 5 release 1.4.3
Using libconfiguration
-L/ms/dist/kerberos/PROJ/mitkrb5/1.4.3-ms0/.exec/ia32.linux.2.4.lib6/lib
-Wl,-rpath
-Wl,/ms/dist/kerberos/PROJ/mitkrb5/1.4.3-ms0/.exec/ia32.linux.2.4.lib6/lib
-lgssapi_krb5 -lkrb5 -lk5crypto -lkrb5support -lcom_err -lresolv
Setting MIT defines
using own defines
Checking if your kit is complete...
Looks good
Unrecognized argument in LIBS ignored: '-Wl,-rpath'
Unrecognized argument in LIBS ignored: '-Wl,/ms/dist/kerberos/PROJ/mitkrb5/1.4.3-ms0/.exec/ia32.linux.2.4.lib6/lib'
Writing Makefile for GSSAPI
Which seems quite reasonable. Except that I want to use 1.4-lib-prod,
which is a symlink to the actual version 1.4.3-ms0. This symlink can be
changed at a later date, so I don't have to recompile the perl GSSAPI
module if I want to address a bug for instance. The krb5-config is too
specific (I could change that ofcourse). This is debatable and I can
agree it is just my problem to deal with.
The second problem is more of an issue, the ignoring if the -Wl
arguments. These are needed at link time, otherwise I don't get a proper
runpath. I can set these using LDDLFLAGS as well as LIBS in
WriteMakefile.
So I propose a mechanism that allows me (and probably other control
freaks) to set the LIBS, INC and LDDLFLAGS directly. See attached patch.
Any objections?
Cheers,
--
Merijn Broeren | Sometime in the middle ages, God got fed up with us
Software Geek | and put earth at sol.milky-way.univ in his kill-file.
| Pray all you want, it just gets junked.
|
|
From: Achim G. <per...@gr...> - 2006-02-06 19:53:52
|
I have uploaded 0.14.
I think that this is a first "make footprint" release,
and more releases have to follow
from Changes file:
0.14 Mo Feb 6 18:46:21 CET 2006
-Maintanace overtaken by Achim Grolms. Thank you Philip!
our base is
<http://perlgssapi.sourceforge.net/>
-deleted gss_mech_set_krb5_v1v2(), gss_mech_set_krb5_v2()
from OID__Set.xs. It makes the compile fail
-deleted gss_mech_krb5_v2() from OID.xs
-deletd gss_mech_set_krb5_v2, gss_mech_set_krb5_v1v2 from GSSAPI.xs
-gss_str_to_oid() and gss_oid_to_str()
are not present in Heimdal implementations.
I've ifdefed them to croak.
Testsuite is adjusted to skip them.
-Added patches from Merijn Broeren to typemap file.
this fixes (hopefully)
http://rt.cpan.org/Public/Bug/Display.html?id=5681
Just grep the file for "Merijn", I've marked the sections
-Added destructor to Cred.xs (Thanks Merijn!)
Thank you,
Achim
|
|
From: Achim G. <per...@gr...> - 2006-02-06 16:15:26
|
On Monday 06 February 2006 15:26, Merijn Broeren wrote: > the differences in Context.xs. There are a few issues, the setsvpv bug > and the addition of a DESTROY for the context. It is Cred.xs an not Context.xs svn commit -m "added destructor to xs/Cred.xs. Thanks to Merijn Broeren" Sending xs/Cred.xs Transmitting file data . Committed revision 25. Thank you, Achim |
|
From: Achim G. <per...@gr...> - 2006-02-06 15:30:23
|
On Monday 06 February 2006 15:26, Merijn Broeren wrote: >The biggest fix I want in is the fix I put in rt.cpan.org.=20 I think that's already in the snapshot I send to you. At the moment I don't anderstand how the black magic in the typemap file works. If it is possible to explain to stupids like me, feel free to send an explanation that I can 1. add to the documentation 2. add to my brain ;-) > and the addition of a DESTROY for the context. Corect. In my implementation I'm using the the return values of gss_release_cred(&minor, &cred);. =46rom my point of view the constructor should warn or die if something goes wrong with gss_release_cred() just to indicate that the using Application has real problems... What do you think? |
|
From: Merijn B. <me...@il...> - 2006-02-06 14:28:02
|
Quoting Achim Grolms (per...@gr...):
> OK, PAUSE account is ready and
> GSSAPI namespace registered (transfered from Philip Guenter).
>
> Now we can start uploading to CPAN :-)
>
> what things *must* be fixed before I start the first upload?
>
Right, I should attach things...
By the way, just have a look through the diff and comment on it. I'll
provide a proper diff against your 0.14 prerelease after I get it all to
compile and tested it.
Cheers,
--
Merijn Broeren | Sometime in the middle ages, God got fed up with us
Software Geek | and put earth at sol.milky-way.univ in his kill-file.
| Pray all you want, it just gets junked.
|
|
From: Merijn B. <me...@il...> - 2006-02-06 14:26:41
|
Quoting Achim Grolms (per...@gr...): > OK, PAUSE account is ready and > GSSAPI namespace registered (transfered from Philip Guenter). > > Now we can start uploading to CPAN :-) > > what things *must* be fixed before I start the first upload? > The biggest fix I want in is the fix I put in rt.cpan.org. Basically all the differences in Context.xs. There are a few issues, the setsvpv bug and the addition of a DESTROY for the context. I've attached a full unified diff between my code and the 0.14 prelease Achim has so thoughtfully send to me. The other big change is that in my version we dropped outright all the mech functions. I see your version also drops two. I don't remember why we dropped them, I'll dig around in my archive, but for now I say go with the deletion. :-) Cheers, -- Merijn Broeren | Sometime in the middle ages, God got fed up with us Software Geek | and put earth at sol.milky-way.univ in his kill-file. | Pray all you want, it just gets junked. |
|
From: Leif J. <le...@it...> - 2006-02-06 07:15:00
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Achim Grolms wrote: > OK, PAUSE account is ready and > GSSAPI namespace registered (transfered from Philip Guenter). > > Now we can start uploading to CPAN :-) > > what things *must* be fixed before I start the first upload? > I suggest you upload now. If you have something which compiles with MIT and Heimdal you have an improvement from 0.13 already. MVH leifj -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFD5vds8Jx8FtbMZncRAhTmAKCiLS7Vrp/7pb/YOQK/lAICFG78FACglrGv 5dyozTQWDr+XFCnKcEjFGPA= =5FJL -----END PGP SIGNATURE----- |
|
From: Achim G. <per...@gr...> - 2006-02-05 23:15:23
|
OK, PAUSE account is ready and GSSAPI namespace registered (transfered from Philip Guenter). Now we can start uploading to CPAN :-) what things *must* be fixed before I start the first upload? Achim ---------- Forwarded Message ---------- Subject: Re: Uploading Modules to other peoples namespace Date: Sunday 05 February 2006 23:18 From: Philip Guenther <gue...@se...> To: ac...@gr... Achim Grolms <ac...@gr...> writes: >In PAUSE you are able to change the Permissions of GSSAPI namespace. >Can you register me as Co-Maintainer or Maintainer? (AGROLMS) Done. Have fun! Philip Guenther ------------------------------------------------------- |
|
From: Achim G. <per...@gr...> - 2006-02-01 19:16:42
|
This is just a test. I hope we will produce good and useful software! Achim |
|
From: Achim G. <ac...@gr...> - 2006-02-01 18:53:47
|
This is just a test. I hope we will produce good and useful software! Achim |
1 message has been excluded from this view by a project administrator.
