[NeXus-tech] NXimpatient

[Kienzle, Paul A.]

>> [Paul Kienzle]
>> 2. IDL is more involved in that it only provides the low level HDF interface.  We could:
>> 
>> (a) show the raw HDF calls from IDL, much like the C example,
>> (b) have users install nexus and use the napi idl interface, or
>> (c) extend Mark's pure IDL wrapper with a similar high level interface as matlab and link to the .pro file from the downloads page.

> [Mark Koennecke]
> This is outdated as it exists only for HDF-4. So a) or b) are the options without much work. In some
> stage we decided to step back from reimplementing the NAPI in another language (IDL) and rather
> use a binding to the C-API. The main reason was maintenance load.

IDL experts can correct me on this, but it seems the current IDL bindings need to be recompiled for the specific IDL installation, which requires that the entire compiler toolchain be available from IDL.  I don't believe IDL ships a C compiler with their app.   This makes it much harder for IDL users to share code that uses NeXus.

We could rewrite the IDL bindings using CALL_EXTERNAL (similar to what we do with python), since that would only require that the user install the NeXus package.  That's better than the current situation.

Richard Azuah has napi45.pro, which he believed was written by Mark  (see attached).  It looks like it supports HDF 5 but I haven't tested it.  This could be updated to the current standard.

It seems this decision should be up to the IDL user community --- whatever bindings they provide, we make them available from the nexus website.

  - Paul
-------------- next part --------------
A non-text attachment was scrubbed...
Name: napi45.pro
Type: application/octet-stream
Size: 30105 bytes
Desc: napi45.pro
URL: <http://lists.nexusformat.org/mailman/private/nexus-tech/attachments/20111027/e7492ad3/attachment-0001.obj>