Hi Ariel,
> So I'd like an [EMAIL PROTECTED]
> ...
Okay, I try to remember 'til Monday to carry the question/suggestion to
[EMAIL PROTECTED] (which would be a more natural place) ...
Recently, I find myself more and more attracted by the idea that we in
fact do way too little for our API users (aka "developers"), and that
fighting for a change here might be worth my previous time :)
> * developers should be "forced" to make their interfaces public in this
> mailing list ("forced" could mean this may be a prerequisite to include
> a CWS in the main stream)
Well, there already is a rule to announce interface changes of any kind.
The fact that this rule is rarely followed is nothing you can change by
"force", but only by education. (Yes, I believe in the truth of "There
is no technical solution to a social problem.".) IoW, we need to raise
awareness of the need to announce API changes.
> * once made public everyone interested can check their correctness
> before being included [2]
That's what [EMAIL PROTECTED] is for (with the same problem of being
an all-in-one list). However, discussing every (in particular complex)
interface/change there is something which is not even required by rules,
since this wouldn't scale. No surprise it's used even less than
interface-announce.
Finding errors in the API before it's integrated / finally implemented
is something which is hard to reach, IMO. I'd already be satisfied if
every API would be *announced*, as this would at least give a
possibility to know about, and potentially discuss, about it. If there
are flaws found in the API after integration, then there's usually also
a chance to change it before the next release.
> * everyone can check that they are *actually* documented, by including
> them in the IDL [3] (as actually there are services/interfaces NOT
> documented but implemented)
Well, I'd expect every announced change to be documented, this is
self-evident, IMO. Or, the other way round: I wouldn't require to
announce non-documented API, since by definition, it's not for public use.
>> something like [EMAIL PROTECTED], or even only
>> [EMAIL PROTECTED] Do you think that'd be useful (if
>> properly fed, of course)?
>
> I vote for [EMAIL PROTECTED]
> And the actual interface-announce should be changed to something like
> internal-cpp-or-whatever-interface-announce so that we all don't get
> confused anymore.
Well, replacing it by an [EMAIL PROTECTED] might be
more suitable.
Anyway. I fear everything said here is something I should have said in
the to-come thread in [EMAIL PROTECTED] See you there.
> [3] That services/interfaces are implemented but not documented is not
> only the developer's fault, it ALSO *our* fault, API clients. The
> example I gave above, is a good case: com.sun.star.xml.sax.Parser has
> been used for a long time (all quotes go back to DannyB example in
> http://www.oooforum.org/forum/viewtopic.phtml?t=4907 form 2004!), but
> nobody ever filed an issue
Well, I it's not really fun submitting issues for all the wrong/missing
documentation ... I'm too lazy for this myself, and I know for sure
there could be many issues ... I tend to think the problem should be
solved at the root, by raising the awareness for the importance of our API.
Ciao
Frank
--
- Frank Schönheit, Software Engineer [EMAIL PROTECTED] -
- Sun Microsystems http://www.sun.com/staroffice -
- OpenOffice.org Base http://dba.openoffice.org -
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
