optimize Namespace

[sqlalchemy-bot]

Migrated issue, originally created by Michael Bayer (@zzzeek)

its time to get around to this.

1. the "file" and "module" arguments to <%namespace> become mutually exclusive. one or the other.

2. Namespace breaks into TemplateNamespace, ModuleNamespace, and Namespace which still is used if there's no template or module.

3. __getattr__ can now be broken down to not require the "has_def()" and "hasattr()" calls, just catch attributeerror and re-raise.

4. more importantly, __getattr__ should put the resolved method into __dict__.

5. when we do the lambda *args, **kwargs:callable_(context, *args, **kwargs), lets use functools.partial . its native

6. inline _get_star(), plus the partial, should speed it up a bit

[sqlalchemy-bot]

Michael Bayer (@zzzeek) wrote:

the above more or less in 717d8a8 cff8367 cb9ebd2. can't really lose the has_def() and hasattr() calls. some runsnakerun profiling revealed the biggest hit as not using StringIO anymore (4737c26) and to get rid of that builtin copy (6c06a32). we're within 10% of jinja2 in that particular test now.

[sqlalchemy-bot]

Changes by Michael Bayer (@zzzeek):

• changed status to closed