You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Kevin Pilch-Bisson <ke...@pilch-bisson.net> on 2002/09/16 17:02:50 UTC
Should we encourage --disable-shared?
Sander and I were discussing this on IRC today, and came to the conclusion
that we should probably stop encouraging people to use --disable-shared.
The only real reason for it is to make setting breakpoints (a little bit)
easier. Hopefully, by this time most of our users won't have to actually gdb
into the svn client.
Anyway, Sander said he could take care of it, but had to go away, so this is
mostly just a reminder for him :)
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Kevin Pilch-Bisson
kevin@pilch-bisson.net
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
RE: Should we encourage --disable-shared?
Posted by Sander Striker <st...@apache.org>.
> From: Branko Cibej [mailto:brane@xbc.nu]
> Sent: 22 September 2002 19:25
> Karl Fogel wrote:
>
> >Kevin Pilch-Bisson <ke...@pilch-bisson.net> writes:
> >
> >
> >>Sander and I were discussing this on IRC today, and came to the conclusion
> >>that we should probably stop encouraging people to use --disable-shared.
> >>
> >>The only real reason for it is to make setting breakpoints (a little bit)
> >>easier. Hopefully, by this time most of our users won't have to actually gdb
> >>into the svn client.
> >>
> >>
> >
> >Hmmm, also "make check" can't be run before "make install" without
> >--disable-shared, right?
> >
> Nope. Testing a shared build with in-tree libs works just fine, I've
> been sending automagic test reports for some months now.
Ben took care of INSTALL in 3199. Thanks Ben! [it was on my plate, but I hadn't
gotten round to it yet].
Sabder
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Branko Čibej <br...@xbc.nu>.
Karl Fogel wrote:
>Kevin Pilch-Bisson <ke...@pilch-bisson.net> writes:
>
>
>>Sander and I were discussing this on IRC today, and came to the conclusion
>>that we should probably stop encouraging people to use --disable-shared.
>>
>>The only real reason for it is to make setting breakpoints (a little bit)
>>easier. Hopefully, by this time most of our users won't have to actually gdb
>>into the svn client.
>>
>>
>
>Hmmm, also "make check" can't be run before "make install" without
>--disable-shared, right?
>
Nope. Testing a shared build with in-tree libs works just fine, I've
been sending automagic test reports for some months now.
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Kevin Pilch-Bisson <ke...@pilch-bisson.net>.
Quoting Karl Fogel <kf...@newton.ch.collab.net>:
> Kevin Pilch-Bisson <ke...@pilch-bisson.net> writes:
> > Sander and I were discussing this on IRC today, and came to the conclusion
>
> > that we should probably stop encouraging people to use --disable-shared.
> >
> > The only real reason for it is to make setting breakpoints (a little bit)
>
> > easier. Hopefully, by this time most of our users won't have to actually
> gdb
> > into the svn client.
>
> Hmmm, also "make check" can't be run before "make install" without
> --disable-shared, right?
>
Actually with the libtool we require, it should be possible to use the in tree
versions, as long as you don't have LD_LIBRARY_PATH set to your installed libs
(which would give them preference over the in tree ones).
> Not sure this is a showstopper, just wanted to point it out. I'm
> still okay with the change, as long as the instructions also point out
> the deal with "make check".
>
> > Anyway, Sander said he could take care of it, but had to go away, so this
> is
> > mostly just a reminder for him :)
>
> Thanks Sander :-).
>
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Kevin Pilch-Bisson
kevin@pilch-bisson.net
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Karl Fogel <kf...@newton.ch.collab.net> writes:
> > If you look in my subversion.spec file, I routinely run "make check"
> > before "make install" and also don't use the "--disable-shared".
> >
> > Just another data point.
>
> Oh, that's the expected situation. But it's the reverse we're
> considering: If you *do* pass --disable-shared, can you run check
> before install?
Wow. Time for me to go home, I can't read OR write English anymore.
Thanks for the data point, David.
-K
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Greg Stein <gs...@lyra.org> writes:
> Sure. 'make check' works regardless of how you build. Ben was doing it all
> the time before he discovered that an LD_LIBRARY_PATH was monkeying up his
> shared build.
If we just document that, then yeah, let's remove the --disable-shared
advice from the bootstrap process.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Greg Stein <gs...@lyra.org>.
On Tue, Sep 17, 2002 at 05:19:09PM -0400, Greg Hudson wrote:
> On Mon, 2002-09-16 at 17:54, Greg Stein wrote:
> > Sure. 'make check' works regardless of how you build. Ben was doing it all
> > the time before he discovered that an LD_LIBRARY_PATH was monkeying up his
> > shared build.
>
> On Linux and some other systems (not Solaris), run paths in the binary
> override LD_LIBRARY_PATH. So until you install, the binaries will (I
> think) choose previously installed libraries over the ones in the build
> tree. "make check" should work if you've never installed Subversion
> before, though.
This is part of the "libtool magic" and why it relinks at install time. The
binaries' rpaths point to the build area for things in the build area. When
you install, then a relink is done to repoint the binaries to the install
area.
(that's the theory; examples to the contrary are welcome -- I'd like to fix
my understanding if it isn't right :-)
Note that the LD_LIBRARY_PATH problem that Ben saw was on FreeBSD.
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Philip Martin <ph...@codematters.co.uk>.
Greg Hudson <gh...@MIT.EDU> writes:
> On Mon, 2002-09-16 at 17:54, Greg Stein wrote:
> > Sure. 'make check' works regardless of how you build. Ben was doing it all
> > the time before he discovered that an LD_LIBRARY_PATH was monkeying up his
> > shared build.
>
> On Linux and some other systems (not Solaris), run paths in the binary
> override LD_LIBRARY_PATH. So until you install, the binaries will (I
> think) choose previously installed libraries over the ones in the build
> tree. "make check" should work if you've never installed Subversion
> before, though.
No, on Linux "make check" always works. We use libtool and that means
that the executable subversion/clients/cmdline/svn is a script that
will relink against the build libraries to produce an executable and
run that. So "make check" uses the build libraries whether or not you
have installed libraries.
I use a shared build on Linux all the time. I use a configure prefix
of /usr/local/subversion and my build produces
$ file subversion/clients/cmdline/svn
subversion/clients/cmdline/svn: Bourne shell script text executable
$ ldd subversion/clients/cmdline/.libs/lt-svn | grep delta
libsvn_delta-1.so.0 => /home/pm/sw/subversion/obj/subversion/libsvn_delta/.libs/libsvn_delta-1.so.0 (0x40086000)
$ ldd subversion/libsvn_delta/.libs/libsvn_delta-1.so | grep subr
libsvn_subr-1.so.0 => /home/pm/sw/subversion/obj/subversion/libsvn_subr/.libs/libsvn_subr-1.so.0 (0x40011000)
--
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Greg Hudson <gh...@MIT.EDU>.
On Mon, 2002-09-16 at 17:54, Greg Stein wrote:
> Sure. 'make check' works regardless of how you build. Ben was doing it all
> the time before he discovered that an LD_LIBRARY_PATH was monkeying up his
> shared build.
On Linux and some other systems (not Solaris), run paths in the binary
override LD_LIBRARY_PATH. So until you install, the binaries will (I
think) choose previously installed libraries over the ones in the build
tree. "make check" should work if you've never installed Subversion
before, though.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Greg Stein <gs...@lyra.org>.
On Mon, Sep 16, 2002 at 04:02:27PM -0500, Karl Fogel wrote:
> David Summers <da...@summersoft.fay.ar.us> writes:
> > > Hmmm, also "make check" can't be run before "make install" without
> > > --disable-shared, right?
> >
> > If you look in my subversion.spec file, I routinely run "make check"
> > before "make install" and also don't use the "--disable-shared".
> >
> > Just another data point.
>
> Oh, that's the expected situation. But it's the reverse we're
> considering: If you *do* pass --disable-shared, can you run check
> before install?
Sure. 'make check' works regardless of how you build. Ben was doing it all
the time before he discovered that an LD_LIBRARY_PATH was monkeying up his
shared build.
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
David Summers <da...@summersoft.fay.ar.us> writes:
> > Hmmm, also "make check" can't be run before "make install" without
> > --disable-shared, right?
>
> If you look in my subversion.spec file, I routinely run "make check"
> before "make install" and also don't use the "--disable-shared".
>
> Just another data point.
Oh, that's the expected situation. But it's the reverse we're
considering: If you *do* pass --disable-shared, can you run check
before install?
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by David Summers <da...@summersoft.fay.ar.us>.
On 16 Sep 2002, Karl Fogel wrote:
> Kevin Pilch-Bisson <ke...@pilch-bisson.net> writes:
> > Sander and I were discussing this on IRC today, and came to the conclusion
> > that we should probably stop encouraging people to use --disable-shared.
> >
> > The only real reason for it is to make setting breakpoints (a little bit)
> > easier. Hopefully, by this time most of our users won't have to actually gdb
> > into the svn client.
>
> Hmmm, also "make check" can't be run before "make install" without
> --disable-shared, right?
If you look in my subversion.spec file, I routinely run "make check"
before "make install" and also don't use the "--disable-shared".
Just another data point.
> Not sure this is a showstopper, just wanted to point it out. I'm
> still okay with the change, as long as the instructions also point out
> the deal with "make check".
>
--
David Wayne Summers "Linux: Because reboots are for upgrades!"
david@summersoft.fay.ar.us PGP Key: http://summersoft.fay.ar.us/~david/pgp.txt
PGP Key fingerprint = C0 E0 4F 50 DD A9 B6 2B 60 A1 31 7E D2 28 6D A8
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Kevin Pilch-Bisson <ke...@pilch-bisson.net> writes:
> Sander and I were discussing this on IRC today, and came to the conclusion
> that we should probably stop encouraging people to use --disable-shared.
>
> The only real reason for it is to make setting breakpoints (a little bit)
> easier. Hopefully, by this time most of our users won't have to actually gdb
> into the svn client.
Hmmm, also "make check" can't be run before "make install" without
--disable-shared, right?
Not sure this is a showstopper, just wanted to point it out. I'm
still okay with the change, as long as the instructions also point out
the deal with "make check".
> Anyway, Sander said he could take care of it, but had to go away, so this is
> mostly just a reminder for him :)
Thanks Sander :-).
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Should we encourage --disable-shared?
Posted by Greg Stein <gs...@lyra.org>.
On Mon, Sep 16, 2002 at 10:02:50AM -0700, Kevin Pilch-Bisson wrote:
> Sander and I were discussing this on IRC today, and came to the conclusion
> that we should probably stop encouraging people to use --disable-shared.
>
> The only real reason for it is to make setting breakpoints (a little bit)
> easier. Hopefully, by this time most of our users won't have to actually gdb
> into the svn client.
++1
I've never had a problem with breakpoints. Just break on main(), enable your
breakpoints (which you prolly set on a previous run), and then continue on.
If that pattern disturbs you, then yes: build static.
But the "default recommendation" shouldn't be a non-shared build.
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org