HBCI-abort

[bock@step.polymtl.ca]

En réponse à Christian Stimming <stimming@tuhh.de>:

> -----BEGIN PGP SIGNED MESSAGE-----
> 
> Benoit,
> 
> On Mittwoch, 11. Dezember 2002 11:18, Eildert Groeneveld wrote:
> > 2. get transaction, no pw required., the importer window comes up (I
> assume
> >     that I got the transactions), then gnucash crashes.
> 
> that crash goes on you. You changed code in the import-export/hbci 
> subdirectory, but haven't tested it AT ALL -- you can't, understandably: HBCI
> 
> stuff can only be used by customers of German banks. Would you please in the
> 
> future refrain from changing code that other users depend on, but where you 
> can't test it at all?!? It turns out you got one FALSE errornously for a 
> TRUE. Well, that can happen to anybody, but I'm quite upset that you 
> introduced it in a part that you didn't test, and you also didn't take care 
> that I review/test it quick enough. 
>
> Why did you want to push your code into mine by all means? Because you wanted
> 
> the HBCI code to use YOUR matcher GUI? By the way, introducing a user 
> preference is a nice idea to force^H^H^H^H^H convince *me* testing your GUI,
> 
> but it wouldn't have been necessary in the first place. I am quite aware of 
> the fact that you are/have been still working on your importer GUI, and I 
> have been looking at this regularly by your provided OFX sample files from 
> libofx. Now why was it necessary to introduce non-trivial code that finally 
> broke the HBCI module for two days?!? Did you think somebody's penis was not
> 
> long enough already?!? If you've had waited some more days (well actually 
> until right now), I would've had tested your code early enough and seen how 
> the GUI is getting into the right direction. If you were so eager to have me
> 
> try this in the HBCI context, you could as well have provided the necessary 
> lines of code as comments in gnc-hbci-gettrans.c and told me which ones to 
> uncomment. Gee. I don't get upset really that often, but this is one 
> occasion. Hrmpf.

WAIT A MINUTE HERE!  I told you, LIVE on IRC that I was about to commit a change
to "Allow HBCI users to chose which matcher they wanted to use" BEFORE I
committed.  You can't claim you didn't see it, since you asked me to specify if
I was talking about the account matcher or transaction matcher.  You didn't ask
me to delay it, and I thought it was implicit you knew I obviously couldn't test it.

Now you accuse me of pushing my code to you (along with references to penis
size) but ALSO accuse me of not pushing you enough to review it.  I asked you a
few times, but I had absolutely no right to push, since on what you spend your
time and how much you spend on it is your business.

The user prefs may or may not have been a good idea, but it still seems logical
to me.  (Not to mention you didn't react at the time).  

Now hadn't never told me before that you have been following my progress on the
matcher thru OFX.  Perhaps if you had we wouldn't be having this discussion. 
But yes, I thought it was absolutely crucial that you take a look at it before
the next release.  An believe it or not, it wasn't because I thought you would
agree to give up you own matcher (In my mind the fork was final for 1.8 so we
both could concentrate on stabilizing and bug fixes).  Most other developers
didn't have time to comment on it (Wilddev took quite a bit of his time to help
me, but warlord and others have far more important work to do right now (fixing
bugs), and the fact that the fork forced me to add this very late in the release
cycle is beginning to make me nervous.  And I don't need my computer engineering
degree to know that a developer is very bad at finding logic errors (TRUE or
FALSE errors are quite typical) in code added on night shifts.  

So hadn't it occurred to you that despite our differences in the past,
differences severe enough to cause a fork, I still considered you the most
qualified person to criticize my work and spot errors?  I don't think any other
active developer spent as much time as you and I did on the design of
transaction importers.  

So yes, I desperately sought your feedback, but NO, it is not the reason I made
the matcher a runtime pref.  As you no doubt saw in the commit, there were quite
a few changes required in many files to allow the two matcher's include files to
be included at the same time.  Had I known you were considering abandoning yours,   
it would have been considerably easier for me to change the source and make it
easy to revert like you did when you first coded your matcher.

Now ok, I screwed up.  I can understand you were annoyed.  Accuse me of writing
crap code in general if you want, it would be unproductive, but I can take it.

But for the rest, I'm sorry but I gave you advance notice, made sure you got the
message before proceeding, and didn't do it right before a beta.  So HBCI was
broken in CVS for two days because I screwed up a true or false and you didn't
have time to look at it immediately.  So what, that's why we have CVS.

All I did is screw up a true or false at 4h30 in the morning.  I don't enjoy
publicly rubbing it in your face when you do.

> Now for the facts: A user preference about which "import matcher" to choose

Thanks for the feedback, I'll read and comment once I have a cooler head.  For
now I am going back to my vacation.

Benoit Grégoire