IDL language

[Lennart Poettering]

On Fri, 08.05.09 11:14, David Zeuthen (david at fubar.dk) wrote:

Heya!

> I think we have seen that this doesn't really work
> 
>  o  No-one really installs introspection XML anywhere. And if they did,
>     it would probably be a dialect anyway.

Avahi installed the introspection files since day 1:

/usr/share/avahi/introspection/AddressResolver.introspect
/usr/share/avahi/introspection/DomainBrowser.introspect
/usr/share/avahi/introspection/EntryGroup.introspect
/usr/share/avahi/introspection/HostNameResolver.introspect
/usr/share/avahi/introspection/RecordBrowser.introspect
/usr/share/avahi/introspection/Server.introspect
/usr/share/avahi/introspection/ServiceBrowser.introspect
/usr/share/avahi/introspection/ServiceResolver.introspect
/usr/share/avahi/introspection/ServiceTypeBrowser.introspect

I really don't see what the great advantage of having an IDL language
would be. Only thing I can see here is that it looks a bit nicer then
the introspection XML. But, uh, if this boils down to just having a
different syntax, then I am all against it.

The introspection format has issues, but XML is extensible and I
believe most of the issues could be fixed inside of the introspection
format instead of creating a new format.

Instead of creating a new language I'd much rather see the current
introspection XML stuff fixed. i.e. integrate the documentation
annotations into the upstream XML format, integrate a way to encode
enums, split up dynamic from static information (i.e. child node info
from static signature data), and a few other things come to my mind.

I fail to see a convincing argument why doing evolutionary changes to
the XML introspection format couldn't bring us the same new features
as your IDL would. Please enlighten me!

Lennart

-- 
Lennart Poettering                        Red Hat, Inc.
lennart [at] poettering [dot] net
http://0pointer.net/lennart/           GnuPG 0x1A015CC4