Rant. was Re: [cvsnt] Re: WSAD 5.0.1 and CVSNT 2.0.35

[Rolf Wilms]

For the following comments I assume that WSAD is based on Eclipse and IBM
didn't patch the Eclipse CVS client to make it incompatible with CVSNT. I'd
appreciate if someone let me know if I'm wrong with this assumption.

The developers working on the Eclipse CVS client can be contacted via this
mailing list:
https://dev.eclipse.org/mailman/listinfo/platform-cvs-dev (creating an
account for posting is required, but this goes quickly)
I'd recommend that you directly talk to them rather than to IBM staff. I've
found them to be very helpful.

The dialog which pops up in Eclipse when CVSNT is detected with alleged
repository prefix is a warning only. If you click yes, the connection will
be established though. I know that the incompatibility between Eclipse and
CVSNT w.r.t. repository prefixes has been a very common pitfall in the past.
I suppose this is why this dialog has been added. Thus I don't see a need
for CVSNT to pretend it was Unix. Just ignoring the warning in Eclipse will
do too. Seems that now that the repository prefix incompatibility has been
solved, this dialog causes as much confusion as it was meant to avoid...

I appreciate your positive attitude towards finding a solution for the
WSAD/Eclipse related problems and I wholeheartedly agree with you that more
effort from third parties is needed to make things work better.


"Tony Hoyle" <tmh at nodomain.org> schrieb im Newsbeitrag
news:c6km9g$ek4$1 at paris.nodomain.org...

> I do this for nothing, and, by and large, enjoy it.  I also get the
support
> of some dedicated people who point out (and sometimes fix) bugs, and
> generally keep my priorities straight.
>
> However I'm continually expected these days to do free techincal support
not
> only for CVSNT but for WSAD.  A program produced by the largest computer
> company in the world (IBM) which costs an arm and a leg and seemingly is
> utterly unsupported.
>
> It took me 6 months to get an answer out of them (a year or two ago now)
> about whether they even were prepared to offer any feedback to me.  They
> said no, and that should have been the end of the matter.  But no, they're
> not even supporting their own paying customers.
>
> I'm sick and tired of answering the same damned questions over and over
> again trying to do free techincal support for a commercial product.  If
IBM
> won't get off their fat arses and support their own software then I
suggest
> you stop buying from them in the future.
>
> Do date I have not received a *single* bug report forwarded to me from the
> WSAD development team.   Around build 57d (2 years ago now) there used to
> be some leakage of repository prefix informaton.  What was their
> response... file a bug report?  Tell me about the problem?  No... post a
> FAQ which effectively says 'CVSNT sucks.  Use Unix'.  This is the attitude
> of 12 year olds not proffesional programmers.  2 years later they are
> *still* insisting there are problems even though those problems haven't
> existed since around that time.
>
> Their latest little game has been to check the version of the server and
> deliberately fail if CVSNT is installed.  Words fail at the utter
> unprofessionalism and sheer childishness of this action.  It really puts
an
> organisation like IBM to shame that they would stoop to such actions.
>
> I put in a workaround for that by lying to the client, but *why the hell
> should I have to*.  *no* other client has these issues.  Where things have
> broken from time to time their developers have filed bug reports and I've
> fixed them usually within a day or two (pushok and smartcvs spring to
mind,
> and of course I get regular updates from tortoise and wincvs developers.
> ViewCVS took some work but with give&take on both sides it's worked out
> OK).
>
> However, I don't believe that the end user should suffer so I'm prepared
to
> work something out.
>
> Currently the state of play is:
>
> 1. There seems to be no technical reason why a CVSNT server with a
> repository prefix + the fake unix option won't work with WSAD.  The output
> is nearly indistinguishable from a 'real' cvshome.org 1.11 server with
this
> setup (running sanity.sh against it should work reasonably well & is worth
> trying at some point).
> 2. I have *no idea* what WSAD is expecting as output, and what it's
parsing.
> I do know it's treating user specific output as authoritative.  Fine, but
> without any idea of what output is significant is *will* break.  Ideally
it
> should be rewritten to use the data from the client/server protocol
> correctly (however I'm not unreasonable and there may be technical reasons
> why this can't be done easily).
> 3. Given (1), rename will almost certainly break WSAD.  I see no way to
> avoid this.
> 4. There is still no point of contact with the WSAD developers.  This
means
> I get no feedback, and there's no dialogue.  Some things (like the case
> sensitive pathname comparison problem) could have been sorted out one way
> or the other long before release.
>
> For the time being I will *not* be answering any more WSAD related support
> queries.  If you want WSAD support your first port of call should be IBM,
> since you've paid for that support.  .
>
> If this is to change someone both (a) running the latest version of CVSNT,
> and (b) running the latest version of WSAD.  Preferably in lots of
> different combinations (with/without prefixes, etc.) and filing bugs with
> enough detail that I can look at it.  Ideally this would be a WSAD
> developer, but given their past support record I doubt this is going to
> happen.
>
> The above person should be prepared to spend time answering WSAD related
> support questions (possibly summarising things on the wiki).  If necessary
> I'm prepared start a support.wsad list to help this.
>
> Tony
>