Tristan Seligmann
Tristan Seligmann
First, a couple of points: 1) The SPI CA is no longer in `ca-certificates` as of version `20151214`. 2) The single PEM CA bundle is only used by software that...
Unfortunately the described feature doesn't actually fully satisfy this feature request; tracebacks are already transformed before logging to Eliot, rather than doing this in Eliot message serialization. See https://github.com/ScatterHQ/eliot/blob/190284e97ab12616ebc6bf8722e2a05a30bde847/eliot/_traceback.py#L86
1. I don't think so, unless it breaks eliottree (but it may be easier to just fix eliottree if it does). 2. It looks like it does. 3. Not sure...
Well, it's only a "leak" in the same sense that it currently is. I guess the concern is that unbuffered log messages may be larger and keep resources like sockets...
Actually I think I was a bit confused about what we were doing. It seems we do already have our own `write_traceback` (actually `writeFailure`) which logs the failure to Twisted's...
Should this be closed now since #179 is merged?
haskell/cabal#4714 is now merged, but I can't tell if it's deployed on Hackage. Any idea where to check on the status of this?
Gah. This can't work exactly as stated, because there's no way to know which cert goes with the private key until you've loaded them with something that can give you...
Note that due to https://twistedmatrix.com/trac/ticket/8227 / https://twistedmatrix.com/trac/ticket/8929 that `abortConnection` call is _never_ made. So Twisted has this bug too :frowning_face:
The code should still _work_, but it will emit a `DeprecationWarning`.