RE: [Firebird-devel] damn gbak Win32 error

[Dmitry Y. <di...@us...>]

Hi all,

Just a few comments of mine (I'll try to be as objective as possible). I
apologize in advance for mixing replies to different people in one post.

DK>>> p.s. people, are we going to release? Or we will wait until Borland
DK>>> will release 6.5 and every user will move to it?
>>
CV>> Going to release with pending issues like this? Are you kidding?
>
DK> Hmm... I'm not saying that we must mark current version as release.
DK> But I don't see any progress moving to release - lot of changes
DK> produces new bugs.

I wouldn't say that new bugs are caused by new features, they're mostly made
while fixing the known bugs. Yes, it's sad but such things sometimes
happen - it's a programming reality.
By the way, I have to agree that the current Beta period is/was too long.
When was it said that RC1 is almost ready to be published? A month ago?

A couple of other thoughts regarding long Beta period:

  - we don't show our timeframes to public (at least I've never seen them),
so users cannot synchronize their business with Firebird development (I mean
stable releases, of course)
  - we shouldn't forget about the competitors (Borland is going to release
IB 6.5 and MySql is releasing their new version with stored procedures,
subselects and other features)

DK> Right now one russian developer have made 5 times much new
DK> features to the engine, making FB work faster and customisable.

Everyone is allowed to start his/her own fork of the source tree. It's okay.
But why doesn't he want to cooperate with either InterBase or Firebird team?

CV>> - If he's so quick and wonderful to make changes, is there any reason
he
CV>> doesn't want to contact Borland or firebird? Is he ashamed of his work?
Or
CV>> does he enjoy violating the license? Or is he doing those changes only
for
CV>> money for a private company that doesn't care to violate the IPL and
retain
CV>> the changes privately?
>
DK> He's doing this for free. FB developers can agree or not with his
changes.
DK> Right now I see EXECUTE VARCHAR feature in the same situation. It is
DK> done, but not yet accepted.

Being a person responsible for this code, I'd like to clarify. It has taken
more time to finish this thing than I initially expected. And I realized
that I'm a bit late when I was reporting about the results of my work very
close to the deadline. So I've asked for the Firebird version to plan this
feature for. In my opinion, it is ready even for v1, but other people
disagree. Is this feature rejected? No, I believe it's just delayed due to
some issues. I've agreed with Ann's suggestion to call the DSQL layer
directly, instead of dealing with Y-valve. But in the same time I don't like
her second objection very much. We'll have time to restart this thread after
v1 is released.

AN>  As far as I know, he couldn't attract your attention to his efforts.
AN> He says that he suggested one time an idea and code for it's achievement
AN> but it was'nt noticed. Some time later it appeared in the FB code but
AN> without mention his name (it can be coincidence, good ideas often come
AN> more than into one head). Don't know, maybe he was'nt persistent enough,
AN> maybe language problems.

I think I could enumerate all guys who are not Firebird members and had/have
got something to contribute. I'll be waiting for Monday to know his name and
I won't be very surprised if I've never heard it. But my memory is not an
ultimate judge and it's possible that he was really suggesting something
useful and wasn't noticed. The similar thing happened a half of year ago
with Alexander Peshkoff's EXECUTE VARCHAR stuff. He was ignored and only
later Claudio has shown some interest, I've contacted Alexander privately
and the stone starts rolling. I understand how hard it should be trying to
join a big group of people who seem to ignore your tries. I think we should
pay more attention to any unknown people who want to cooperate, even if
their suggestions are not going to be definitely accepted. And I don't think
a language barrier is an issue here since we have a lot of people from
different countries who could assist.

AN> BTW, what is the way for developer who made
AN> some noticeable changes into code for own purposes and want this changes
AN> to be included into general stream? I mean not a bug report or feature
AN> request but really noticeable changes.

Let me give you a real example. Yes, it's the same EXECUTE VARCHAR thing ;-)
The neccessary part of the system being developed by Alexander was a dynamic
SQL execution. Such an ability was a primary requirement for a RDBMS that
should be used. So the only way for IB to be used in this project was a
presence of such a feature. Alexander has implemented it over IB 6.01
sources and, since he doesn't want to support the third branch of the code,
he suggested to include it into the IB or FB tree. I've forced him to deal
with FB instead of IB and helped him to port his changes to the FB codebase.
I've also suggested him to ask Mark to get write access to the CVS, but
unfortunately there were permanent problems with his Internet connection, so
he's preferred to allow me to continue his work. So we were working together
polishing up this stuff and after a bit more work this feature will be
finally included into the engine.

SL> In general, I would think that an approach which involves this list at
SL> the start of the process (before coding) would be the most successfull.
SL> A simple posting saying, "I've been thinking about doing this..." (along
SL> with some outline of the approach to be used) would be the best starting
SL> point.

Sounds reasonable. I promise to be at least the one person who reacts ;-).

CV> My "business" (if such a tiny thing can be called business) is to make
CV> custom small applications (mainly with IBO) that I give to my customers
CV> with FB. Why? Because if a bug appears in the engine, I have chances
CV> to find it, I have chances to fix it if I understand why it happens and
CV> finally I have chances to integrate it with the main stream (CVS)
instead
CV> of keeping the fixes for me only.

Claudio, you've described exactly my business and my motivation in the FB
development too ;-) There's no need to say that all of us would very much
like to see FB the most powerful, stable and widely used RDBMS in the OS
world.

CV> Of course it doesn't have. We need more committers. I think Sean has
said
CV> this repeated times and invited people. But we can't know that in
country
CV> ZZZ there's a developer that posted his changes to a small web page in a
CV> regional web space provider, right?
CV> Anyway, what would you have said if we take all the changes from that
CV> unknown for now Russian dev and the engine turns wild? Would you have
flamed
CV> us, too?

You're simplifying things a bit. I believe any developer who suggests
his/her changes is responsible for its quality, but I know there's also the
fact that the complexity of the codebase is high enough to predict/test all
interactions between engine internals without years of experience. That's
the reason why everyone listens to Ann, Jim, Ivan and Dave. But even if that
person have implemented a couple of enhancements and do support them, it's
quite difficult for his/her to provide full QA for the code (a couple of
server architectures, a lot of supported platforms, TCS problems and so on).

To be honest, not being official IB/FB seller/provider/supporter and thus
not having problems with a lot of crying users, I'm quite satisfied with the
way and speed Firebird grows. I've got 0.9.4, 0.9.5 and even 1.0.0 Beta
(!!!) running without a problem in a few production systems. But don't
consider it a flackery <g>, we still have a lot of things to do.

Cheers,
Dmitry