Re: segfault in ruby bindings triggerd by config=Svn::Core::Config.config(@cfgPath)

Hi,

In <CEAB627200FA5C499F8CCB66C26905F30F0C07@wiems02014.ceu.corp.heidelberg.com>
  "segfault in ruby bindings triggerd by config=Svn::Core::Config.config(@cfgPath)" on Tue, 25 Jul 2006 16:42:00 +0200,
  "Husterer, Thomas RD-CP1" <Thomas.Husterer@heidelberg.com> wrote:

> In the ruby script in the Svn::Client::Context object I read the config files
> and assign the result to the local variable 'config':
>
> self.config=Svn::Core::Config.config( @cfgPath )
>
> This works in principle, I can access any kind of server urls. But after a
> short time I get a sporadical segfault wich looks such as this:

I see the problem but I have no time in this week. So I'll fix the problem
next week. Sorry.

> Do I use the wrong method? (i did not find another one)

No. This is a the Ruby bindings' bug.

> Is there really a bug in the pool handling? (I cant comprehend)

Yes.

> PS.
> By the way, I fear that in client.rb the most methods which use
> a 'rev' and a 'peg_rev' parameter have swapped their positions in the
> argument list.
> In the c-header files mostly the peg_rev- comes before the rev-argument
> but in the client.rb this is reversed.
> This is not realy a large problem because the arguments are also swapped
> in the calls to the c-functions.
> Therefore the user has to call the ruby-methods like the c-functions.
> The only wrong effect is the handling of the default values of the parameters
> and the wrong documentation effect.

It is for backward compatibility. The Ruby bindings are develeping before
peg_rev argument exists. So, peg_rev argument was appended to argument
list.

Thanks,

--
kou
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org