20120430
Spent some time experimenting with allowing the
To address both the incoming HTTP socket's filehandle needs to be carried around throughout the current request. This could allow binary content to be written straight to disk from the socket, without going into RAM.
However, getting this to work without making changes throughout
Perhaps a Tiddler subclass could be useful? More poking required.
At the same time I did some analysis on the
text
attribute of a Tiddler
object to be a file
object to see if this might be more efficient when dealing with binary tiddlers. Unfortunately the change does not address the recently raised JSON PUT of binary data issue.To address both the incoming HTTP socket's filehandle needs to be carried around throughout the current request. This could allow binary content to be written straight to disk from the socket, without going into RAM.
However, getting this to work without making changes throughout
StorageInterface
s and Serialization
s is likely hard. Of course that's speculation. It could be spectacularly easy, but the concern is to not impact existing code too much.Perhaps a Tiddler subclass could be useful? More poking required.
At the same time I did some analysis on the
sqlalchemy
code to see if I could locate where problems with PUT
and DELETE
may be coming from. For DELETE
there were issues with loading too many objects into the session. Still unclear on PUT
.Syndicated 2012-04-30 21:50:59 (Updated 2012-04-30 22:58:20) from cdent