Re: Interesting coverage of our project on the SQL-Ledger-users list

["Chris Travers" <..hidden..>]

Just as a note here:


>>
>> A large part of the problem here is that most RDBMS's require
>> licensing fees.  If you start requiring multiple RDBMS's, costs go way
>> up
>
> Its more than that, learning curve goes up for the major contributors
> and you increase the burden of testing, debugging, bug fixing etc...

Actually, I meant requiring for the customer.  Yes this increases the
learning curve, but it also increases licensing costs since these are
usually licensed per user.

If we wanted to support Oracle, DB2, MS SQL, etc. we could get
developer editions etc at little cost.  However, as you point out this
makes it harder for us to produce quality software.
>
>>
>> I don't think there has ever been a willingness on the part of the
>> core team to support MySQL-- that raises data integrity issues we
>> don't even want to try to address.
>
> Let me be real blunt. I quit if we support MySQL. I have zero desire to
> go down that path (or Oracle or otherwise). The only way I would
> consider it, would be if there was a valid technological reason to
> reconsider PostgreSQL as our platform. MySQL does not equate and neither
> does Oracle. People running LedgerSMB don't need RAC :)

Just to be equally blunt:  your opinion is not the only voice on the
subject and if we generally disagreed with the notion then you would
no doubt quit ;-)  But we don't disagree.  The fact is that every
member of the core team backed that decision its merits.  This is and
was a core committee decision based on concensus, not based on
worrying about whether or not CMD would be a part of the project.  And
this needs to be remembered.

My main point is that the decision to support PostgreSQL only is a
decision supported by every core committee member and not something
which is based on one company's interest in the project.    Instead it
is based on the merits of the choice based largely on the criteria I
mentioned (complexity of QA, maintainability of the code, ease of
integration, etc).  In general, moving to PostgreSQL-only is actually
a net win in terms of ease of integration.

If someone wanted to port to Oracle, MS SQL, etc. I would be more than
happy to provide high-level pointers.  However, I would not expect to
provide a great deal of time and effort, and I would not expect it to
be something that would make it into the community distribution.

>
>
>> This means basically the question as to whether or not to try to run
>> on Oracle, DB2, and MS SQL.  In general, I think that porting to
>> Oracle would be remarkably easy because of the fact that the stored
>> procedures are largely in PL/PGSQL which is very similar to Oracle's
>> PL/SQL (both are hybrids of PL/1 and SQL, and if someone wanted to do
>
> But what about plperl :)...

Currently we have avoided relying on PL/Perl.   I dont know of any
current uses of that language in /trunk.

Until I can test PL/Perl on Windows using Vanilla Perl I would not
support relying on it.  Even branching out into multiple procedural
languages is going to make QA much harder.

>  We need to focus on one set of core talents.
> Do that right. If people don't like the fact that we don't run on MySQL,
> then let them fork us.

Not worried about MySQL.  It doesn't really do what is necessary for
an application like this.

I wouldn't provide any assistane to such a fork because there would be
no possibility of future collaboration due to big differences in the
platforms.  This is different than an Oracle fork which could be
sufficiently close that there could be future collaboration regarding
code.

> Seriously, that is what FOSS is all about. I
> applaud anyone who wants to take the energy to do such a thing.
>
I would go one step further:  Any mini-fork where there is a
likelihood of future collaboration I would support and provide advice
and technical assistance to.  However:  1)  We only rely on FOSS.  and
2)  We aim to provide better WIndows support than we started with.


>
> Actually I don't agree with this. People that are going to really want
> to run LSMB are already running FireFox.

For now.  However, I expect the appeal to become broader as the
software gets better.

>  There are of course the
> exceptions but Firefox (which is actually set to hit 20%) is the FOSS
> browser of choice.

I do most of my testing in Epiphany.  Safari is also supported.  I
would assume Opera works.

This is largely a matter again of realizing that the hoops necessary
for IE support were problematic and lead to problems of
maintainability.  Until IE fixes those core issues, I can't see us
offering support.   Now that 8.0 will fix these showstoppers, what
will be necessary is for someone to take the lead and test with IE8
once release candidates are available (I am not going to worry about
accomodating early beta software).  However, I don;t have the time to
do this, so it would have to come from the community.
>
> Not to mention, its all about killing the fly with a cannon. Oh, you
> hate quickbooks? Oh you like LedgerSMB? Well you need Firefox... 6 weeks
> later...

I don't foresee dropping "we recommend Firefox."  However,
StevenMarshall does have a decent point. My company has lost customers
over the IE-is-unsupported issue.  It would be nice to say "Yes, IE8
works, but we still recommend Firefox."

Best Wishes,
Chris Travers