Обсуждение: Various issues

Dear all,

Here are some issues:

1) PostgreSQL installer (wizard)

http://www.networksimplicity.com/openssh has released an OpenSSH minimal
Cygwin installer. They have had 300.000 tracked downloads so far. If true,
this numer is superior to Cygwin downloads itself.

The installer is incompatible with Cygwin (as it installs a minimal Cygwin
layer). They also provide a minimal configuration GUI. Is this a workable
solution?

I got in contact with the installer maintainer to ask the cons/pros of such a
solution v.s. a pure Cygwin layer.

For the moment, I still agree the solution is to create a wizard in pgAdmin2

2) My current work is stopped

Right now, I am too bust to work on pgAdmin2. Feel free to pick-up any issue I
should be working on (if need to have it done quickly):
- Handle two separate connexions (one for schema, another for data queries).
The user should be able to define client_encoding for each connexion.
- Work on PostgreSQL wizard.
- Refine pseudo-alter trigger code (it was removed because unfinished).
- Add pseudo ALTER DROP COLUM (this should not be difficult using pgSchema).
I will be back on the project as soon as possible.

3) KDE3 is marvelous
I now use KDE3 which is a very nice environment. I tried writing some code in
kDevelop. It seems possible to write code fast... Therefore I am still asking
myself ***why***we should continue develop pgAdmin on Windows...

In my humble opinion, KDE needs a real database abstraction layer (like
pgSchema) with a multi-vendor interface. The only solution today is Gnome
libgda. Unfortunately, libgda is not well-written. A good abstraction layer
needs inheritence (C++, not C) and XML to handle specfic features of each
database provider.

Secondly, KDE3, Konqueror and KDevelop would probably welcome a pgAdmin port.
This would immediatly give us thousands of users. So why bother with
Windows?

Thirdly, qt3 applications can be compiled under Windows. And KDE3 is beeing
ported to Windows using Cygwin. You can be sure it is not a year before we
can use KDE3 under WIndows and MacOSX.

Cheers,
Jean-Michel

> -----Original Message-----
> From: Jean-Michel POURE [mailto:jm.poure@freesurf.fr]
> Sent: 12 April 2002 11:42
> To: pgadmin-hackers@postgresql.org
> Subject: [pgadmin-hackers] Various issues
>
>
> Dear all,
>
> Right now, I am too bust to work on pgAdmin2. Feel free to
> pick-up any issue I
> should be working on (if need to have it done quickly):
> - Handle two separate connexions (one for schema, another for
> data queries). The user should be able to define
> client_encoding for each connexion.

Is this really required now? (Considering that multibyte just isn't going to
work with VB6 anyway unless we can get some better controls)

> - Add pseudo ALTER DROP COLUM (this should not be difficult
> using pgSchema). I will be back on the project as soon as possible.

Shouldn't be required. There is much discussion about this on pgsql-hackers
at the moment, and it has been acknowledged that acceptance of patches has
be too strict in the past. I think DROP COLUMN will almost certainly be in
7.3 now.

> 3) KDE3 is marvelous
> I now use KDE3 which is a very nice environment. I tried
> writing some code in kDevelop. It seems possible to write
> code fast... Therefore I am still asking myself ***why***we
> should continue develop pgAdmin on Windows...

I did pose the question recently about a move to C# for primarily this
reason. Most responses indicated that people would *not* be happy if we did
this :-(.

> In my humble opinion, KDE needs a real database abstraction
> layer (like
> pgSchema) with a multi-vendor interface. The only solution
> today is Gnome libgda. Unfortunately, libgda is not
> well-written. A good abstraction layer needs inheritence
> (C++, not C) and XML to handle specfic features of each
> database provider.
>
> Secondly, KDE3, Konqueror and KDevelop would probably welcome
> a pgAdmin port. This would immediatly give us thousands of
> users. So why bother with Windows?
>
> Thirdly, qt3 applications can be compiled under Windows. And
> KDE3 is beeing ported to Windows using Cygwin. You can be
> sure it is not a year before we can use KDE3 under WIndows and MacOSX.

Yes, but qt requires coding in C/C++. If I have to rewrite 30-40K lines of
VB code to C, I can see that becoming more like 100-150K lines. VB does have
limitations, but no-one can deny that it is very good for building
applications quickly. The beauty of C# is:

1) Despite 8 or more years tinkering with C, I already feel more comfortable
with C# after a few weeks.
2) Whilst not quite as rapid as VB, it is definitely faster to build
applications than C/C++.

I very much hear what you're saying Jean-Michel, and agree with the
sentiment, however, primarily it will be me doing the coding I imagine so my
main priority will be ensuring that *I* am comfortable with whatever we
choose.

Regards, Dave.