jmoiron plays the blues

great overview of #Facebook's MySQL tech talk outlines solid approach most scaling efforts independently discover: http://goo.gl/y8sTG

@schinckel maybe https migration growing pains? or was it always that braindead

added a second row of virtual desktops, accessed w/ shift+mousewheel; #flavors requires so much real estate :) working well so far

#19 to argot version 0.6 on pypi

#18 to argot version 0.6, change url to bitbucket, add extras to `argot` script including auto-pygments styling to generated document

released new version of my markdown dialect "argot"; amalgam of moin blocks & md, easy HTML docs w/ syntax highlight http://goo.gl/Oe8LD

wrote a reply blog post to the recent "virtualenv tips" by @ericholscher http://bit.ly/dvgxJI

ce6e75f6 to iris change '--force' to '--yes' for flush (so iris flush -y now to flush), add '-c' to list to get a quick count of how many photos are in iris' db, implenet iris add -r (recursive), catch a bug in the exif parser for .THM files

dd895da8 to iris reorganize utils.py, add 'recursive_walk' function which presents us with a list of full paths given a list of paths that are to be recursively walked (perhaps a list of directories, or directories & files mixed)

94b06d84 to iris added --profiler and --timer to the generic options (mostly for development), but also added a --parallelize switch (which I might make default in the future) which seems to really improve performance: http://jmoiron.net/~jmoiron/multiproc.png

@jacobian johnny-cache, but of stuff I haven't written, django-extensions or django-debug-toolbar

@mitsuhiko that was my point in 'github has won' but people basically told me to go fuck myself

found strange crash bug in #python #mongodb driver while being naughty: http://jira.mongodb.org/browse…

ed30b105 to iris fix list to use the a paged auto-classing query

3803279c to iris added flush command, moved OpenStruct back to inheriting from object to avoid pymongo bug, fix add command

81d957df to iris fix paging cursor to allow limits, sorts, skips of their own, independent of the internally managed values used for paging

676e9b35 to iris get a working paging cursor, mostly useful to keep 'sync' operations in a "constant" bound memory space despite the size of the image collection

f7272e94 to iris rename BulkUpdater -> BulkInsterter, as collection.update has some other connotations I don't want to overload

15b4dc63 to iris add new BulkUpdater backend class, which creates an updater that will intelligently mass-insert, to a threshold of 100 documents. even if they are all truly updates, it still saves N reads (N=threshold) per flush as normally these documents would have to be checked for existence in the collection manually by their unique value, but here it's done in one query

I feel like #ubuntu times their xulrunner updates for whenever I've compiled #mongodb from source...

73ff7c6f to iris wip for new backend work with thread-safe batching of mongo queries for doing large imports in a fixed(ish) amt of memory maintaining speed

1b329df1 to iris fix the file loader not getting the exif keys quite right for filtering

4735e67c to iris have sync & list react to files that have been deleted (or return)

94136a0e to iris add a sync command that resyncs everything in the database and marks as 'gone' if they have been moved or deleted

e9e75316 to iris get add & list commands working for simple images