Re: when will HDF5 support Unicode?

Quincey Koziol <koziol@xxxxxxxxxxxxx> writes:

>> > When will Unicode be supported in HDF5?
>> > Ed
>> > 
>> 
>     We are waving around the term "Unicode" like we know what it means and I'm
> pretty certain it is more complex than we think.  This article gives a
> reasonable overview from a software developer's position:
>     http://www.joelonsoftware.com/articles/Unicode.html
>
>     We should decide which character sets we are going to try to support and
> then find some open source software which performs conversions between ASCII
> and Unicode strings - I don't think we want to be in that business.
>
>     Quincey
>
>

Obviously this is a complex issue that you guys need to consider in
terms of HDF5. I don't think we have a strong enough requirement on
this end to drive it.

Specifically, HDF5 1.8 is only going to handle ASCII, correct?

If so, then I think the netCDF-4 approach would be to support it when
you support it in HDF5. As Quincey points out, the requirement is
vague.

I think what I really intended when I wrote the requirement was that
scientists should be able to use netCDF to generate datasets, which
would then be displayed in some way. It would be nice if the names
displayed on the scientists' graphs could be in their native language,
if they so choose. This isn't possible with ASCII-only strings.

However I don't hear anyone else pushing for this requirement, and I
don't want to implement it on top of HDF5, I want HDF5 to implement it
and netCDF-4 to just lazily use it!

So I will drop it from the netCDF 4.0 requirements and hope for it in
some future version of HDF5.

Ed


-- 
Ed Hartnett  -- ed@xxxxxxxxxxxxxxxx