"Alvaro Herrera" <alvherre@commandprompt.com> writes:
> Bruce Momjian wrote:
>
>> Based on that reaction I am not going to bother uploading my copy of the
>> typedefs.
>
> Please reconsider. Not having pgindent work at all is not better than
> it working "only" 98%.
I think I'm rescinding my objection to checking a canonical list of typedefs
into CVS. I didn't realize how hard it was to generate these typedefs or how
important it was to have everyone using the same version.
Since we really *don't* want individual developers rebuilding the list of
typedefs we don't have to worry about conflicts when the upstream list
changes.
Bruce's list of typedefs seems like a good start point for a "canonical" list
of typedefs. The idea of gathering the lists from the build farm and
consolidating them sounds like a good plan going forward.
The only thing is that if the whole point is to have patch submitters run
pgindent on their own added code it won't work since their own code will be
precisely the code with the missing typedefs. How easy is it to manually add a
handful of typedefs to the list?
-- Gregory Stark EnterpriseDB http://www.enterprisedb.com Ask me about EnterpriseDB's RemoteDBA services!