[Gdal-dev] Re: GDAL's Python Identity Crisis (was Re: Import Gdal)

[Lucena, Ivan]

Hi There,

I am very happy that you guys are going to work on the GDAL's Python 
extension. I just want to add my humble comment for why I see it as very 
important:

Some time ago, I used GDAL 1.3.2 with Python to develop a import routine 
for data from the IPCC archives. That thing attract so much attention 
from GIS and Remote Sense students that I decide to build a installer 
with InnoSetup and give it away. They just needed to run my GDAL 
installer, Numeric, Python and PythonWin installer.

I am not going to comment on the current situation, but...

The point I want to make is that this combination of scripting language 
and GDAL is a very powerful tool for students to put their hands on the 
core of GIS technology in a very simple and efficient way. Not just 
reading about it or running somebody else's code.

That is what I think: They are not programmers, so we weed to make it 
easier for then. They use Windows and they need the PythonWin IDE 
because it has code completion and interactive debugging. After typing 
"gdal." they need to see what is available. They need to run step by 
step to check for their mistakes and to learn how that "programming 
stuff" works (sort of).

And I believe they will have it.

In a week!

Best regards,

Ivan


Frank Warmerdam wrote:
> Brian Hamlin wrote:
>> On Sep 20, 2007, at 9:59 PM, Christopher Barker wrote:
>>>
>>>> I think we should only do this for the next-gen bindings,
>>>
>>> I agree, though it might be better from a confusion stand point to 
>>> have them both the same -- I'm really not sure.
>>
>> Why not  leave the old name space with the old Python bindings, and 
>> preserve the old test suite in a folder with those names, all as 
>> legacy. Then the new namespace goes with the new bindings and (perhaps 
>> the same) tests but in a new folder with the new invocation. Sounds 
>> like a clean setup.
> 
> Brian,
> 
> No, this would not be acceptable.  For the near future old and new
> bindings will be supported, and the test suite has to be able to run
> against both.  It is acceptable for the test suite to skip specific tests
> when run against old or new bindings but most of it should continue to
> work with both for at least 1.5.
> 
> (BTW, when I say unacceptable, I mean I would vote -1 on such a proposal.
> I'm not the final word obviously - I could be overridden)
> 
>> That way if you ever look at a test file, you know for certain which 
>> generation you are looking at.
>> ps- extra bonus, do that for 1.5 and leave 1.4.3 working as it stands :)
>>
>> caveat - I dont use Python much
> 
> I use python every day, and write and maintain many of the test scripts.
> 
> Best regards,