Re: [pygccxml-development] Future of pypp_api...

[Roman Y. <rom...@gm...>]

On 8/20/06, Matthias Baas <ba...@ir...> wrote:
> The last modification that broke pypp_api was the transfer of the
> "write_main" parameter from multiple_files_t.write() to the constructor.

The person who added( me ) this flag to the "write" method completely
missed the fact, that
method "write" should be implemented as is, without adding additional arguments.
Otherwise, it does not implements "file_writer" interface. Everyone
who uses Py++
module_builder_t class had nothing to do with this change.

> We already had a few similar changes before.
> Then, I think it was the call policy objects that have been moved to a
> different module so that they had to be imported from somewhere else.

That happened, when we just started to work on high level API, so this does
not count. Also, in ALL cases, Py++ and pygccxml packages have __init__.py
file, that contains\\introduce all references to the functionality
provided by package.

> Well, these changes are no big deal (especially as py++ still hasn't
> reached v1.0) and for me, all those modifications were "shielded" by
> pypp_api and I never had to update my main script so far, so I don't
> really complain about this... (but as you were asking)... ;)

This is good.

Users of module_builder_t API are also protected from some kind of
changes, exactly as in your case.

-- 
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/