directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Emmanuel Lécharny <elecha...@gmail.com>
Subject Re: [VOTE] Apache LDAP API 1.0.0-M32 release
Date Mon, 19 Oct 2015 08:52:48 GMT
Le 19/10/15 10:30, Radovan Semancik a écrit :
> On 10/19/2015 02:23 AM, Emmanuel Lécharny wrote:
>> I think we can wait for another release, that may come quite quickly
>> (I have myself some additional fixes for the LdifAnonymizer).
>
> OK. No problem. Just let me know when the M32 release is done so I can
> commit my code.

I'll close the vote tonite, afte rmy day job ;-)
>
>> Regarding the missing controls/extOps, here is what I would suggest :
>> we could spend some time implementing a batch of the missing AD
>> elements, and cut a release as soon as it's done.
>
> Hmm, but there is a problem. It does not make much sense to implement
> the controls "theoretically". They have to be tested with real AD
> instances. We all know how AD "works", eh?

Meh... So right. OTOH, there is nothing forbidden us to get the code
added, with internal tests, for someone to test it against AD.
>
> So I would counter-propose to implement the controls gradually when
> someone needs them and someone is able to test them. This is the case
> for the "Deleted" control: I need it and I have a setup to test it
> with real AD instance. Yes, this specific control is trivial. But
> there are quite complex controls in the AD set ...

I have yet checked teh syntax for all of them.

What I would suggest here is to create either one JIRA for all the AD
controls/extended, or one JIRA per control/extended. The second option
might be a better solution, even if that mean 50 JIRA will be created
(OTOH, closing them fast wikll improve our opened/closed ration in a
nice way ;-)
>
>> For controls, it's not necessarily complex, it's just a bit time
>> consuming (especially the tests).
>
> Well, yes, unit tests are a bit time consuming. But tests with real AD
> are insanely time consuming as AD will not tell you what is the
> problem. So, unless you already have a setup where you can test it
> easily I see no point in implementing something that will not be tested.
I don't have any AD close to me, so I have to trust you on that :-)

>
>> The only part I'm not sure of is which ones should we include and
>> which ones should we ignore. I suspect we should go to the full
>> extent and make the API as complete as possible... 
>
> For me it is the "Deleted" control now. Maybe SD_FLAGS a bit later.

Ok, let's go for the first one quickly then.


Mime
View raw message