[Mono-devel-list] Tests for System.DirectoryServices

[Sébastien Pouliot]

Hello Boris,

[As far as I'm concerned] as long as any user (without an LDAP server) can
"make run-test" in the System.DirectoryServices directory without seeing any
errors (e.g. running a subset of the tests) and without any big performance
impacty (e.g. multiple timeouts) then I have no problem with it.

Sebastien Pouliot
home: spouliot at videotron.ca
blog: http://pages.infinit.net/ctech/poupou.html


> -----Original Message-----
> From: Boris Kirzner [mailto:borisk at mainsoft.com]
> Sent: 19 juin 2005 11:27
> To: spouliot at videotron.ca
> Cc: kostat >> Konstantin Triger; mono-devel-list at lists.ximian.com
> Subject: Re: [Mono-devel-list] Tests for System.DirectoryServices
>
>
> Hello Sébastien,
> I've finished implementing the default ldap server feature (currently
> waiting for devlist approval) and I've also made an example for defining
> default ldap server configuration.(attached are an example file and the
> proposed patch for the case you missed my mail to the devlist).
>
> The main point if the change is: if user does not specifies a path of an
> entry, the entry should point to default ldap server on the network
> (actually, to default naming context of the default ldap server). The
> default naming context can be retrieved from the server rootDse entry,
> so the only information we need to provide a desired functionality is a
> default ldap server host and port.
>
> User does not have to provide his own configuration: the default
> configuration (absence of user configuration file) means "use
> localhost:389". User _can_ overwrite this with his own values using app
> config or machine config.
>
> Is the "root" System.DirectoryServices is the best place to place an
> example and the explanation file?
> Should also user configuration file reside there (since, for example,
> the tests run by default in this directory).
>
> The second point is that DirectoryServices tests require environment
> variable of (some, not necessary the default) ldap server to run on.
> So, the tests logic can possibly be: if no environment variable present,
> test prints warning message and silently passes(or, maybe, fails?). If
> someone wants to run this tests, he should define both environment
> variable and a user configuration file (or configure the host running
> these tests to act also as a default ldap server)
>
> What do you think about this?
>
> Thanks,
> Boris
>
> Sébastien Pouliot wrote:
>
> >Hello Boris,
> >
> >
> >
> >>I understand that use of machine.config or app.config is far from ideal,
> >>but at least it can provide us with a basic solution.
> >>
> >>
> >
> >When things gets too complicated people quit. I don't think many
> people are
> >gonna be runned the LDAP tests but I wouldn't want to discourage
> anyone. So
> >the easier the better ;-)
> >
> >Why don't keep the environment variable (very simple) and point it to a
> >configuration file (XML or plain ASCII) ?
> >
> >That would mean:
> >
> >* that, by default, all non-network tests in System.DirectoryServices.dll
> >will be executed (here I'm thinking about the few permissions
> tests already
> >present _and_ future CAS tests);
> >
> >* it would be possible to check into SVN a configuration sample file that
> >any user can copy (so it won't get in conflict with SVN) and edit;
> >
> >* it would be possible to have multiple configuration files, e.g. if you
> >want to test against different LDAP servers (machine or software);
> >
> >* all configuration could be done in a single place whatever the
> >runtime/version you are using.
> >
> >
> >
> >>Do you think there is a better way to provide a "default" behavior,
> >>either at DirectoryServices or Ldap level, such that it will be easily
> >>configured for each client application?
> >>The trivial solution - default is "LDAP://localhost:389", does not looks
> >>satisfactory and, in addition, does not provides a solution for
> base dsn.
> >>
> >>
> >
> >Without a valid configuration file I would simply ignore all tests
> >(requiring a server). This is similar to what I'm doing for the
> CAS tests.
> >If someone doesn't exclude them (default for "make run-test") and doesn't
> >supply --security to the mono runtime, then they are all ignore
> (and nicely
> >nunit still shows them as ignored when they aren't excluded).
> >
> >Sebastien Pouliot
> >home: spouliot at videotron.ca
> >blog: http://pages.infinit.net/ctech/poupou.html
> >
> >
> >
>
> --
> Boris Kirzner
> Mainsoft Corporation
> http://www.mainsoft.com
>
>