#pyramid

the python csv module has done a pretty good job for me of creating double-clickable csv files that import correctly into excel

even with columns of arbitrary text... the main issue is excel's type-guessing

and also excel being a pile of rickety shit.

non-ascii characters are unreliable at best

raydeo: only fine as long as you only have ASCII

who would use anything else?

And Excel will fuck anything that looks like a number but isn't

yeah that last one is what I've hit repeatedly

Good question, only fools.

openpyxl used to do default to guessing. Not only is it very error prone, it's also very slow.

Personally, I think something like HD5 is what people should use for moving data around. But as long as there is a market that insists on using Excel…

Someone recently asked for support for files > 4GB!!!!

Like social media it's probably proof of our decadence.

Anyway, got to to go and give my lightning talks.

I just noticed that request.json_body - is not cached

is that on purpose?

i assumed that it would be reified

Ergo: for the discussion on that: https://github.com/Pylons/pyramid/pull/1014

We just have an `add_request_method` that makes `cached_json` reified

sontek: yeah I was expecting this would be the reasoning for it to be this way

and probably will be doing it same way

sontek: I had some quite nice ziggurat releases lately :-)

it could be cached if we were smarter about it

raydeo: doesnt have to be, for some reason i just assumed upfront its cached

we'd have to store a key in the environ related to the body and see if the body changed later

if we weren't smart about it though it could end up storing a ref to an old body

raydeo: mcdonc should I bother resolving the merge conflicts in https://github.com/Pylons/pyramid/pull/1570 ?