rename call methods to transform and postprocess #4044
Conversation
|
Thanks for the ping. This is not a 1.0 bug fix, but a 2.0 feature, correct? |
|
No problems here... ;-) |
|
yes, should be 2.0 |
|
+1. Want to keep call as deprecated alias ? |
|
@Carreau I consulted with Min about that as I was writing the PR, and since we said nbconvert is alpha-tech preview, we don't need to - we're allowed to change things there at will (it would be kind of a nightmare to keep everything compatible with the code we have in 1.0, since so much more cleanup will go into nbconvert for 2.0 |
|
@ivanov ah, yes, you are right. |
|
possible change: for symmetry / clarity, should we rename 'transformers' to 'preprocessors'? It seems more descriptive, and we have frequent confusion about the difference between transformers and filters (from those that aren't aware that filters are a jinja thing). |
|
I'm +1 on the preprocessor renaming |
|
yes, I agree... |
|
just to clarify, I'm adding to the PR a global search-replace of Transformer -> Preprocessor, yes? |
|
Sure - also probably means change |
|
yeap, ok, that's what I'm doing, update coming shortly |
|
Note that this will conflict with #3747 |
|
done with preprocessing, I'll send you a patch to fixup 3747, Matthias, if you'd like |
|
No I can do it, just want to pint out. |
|
have you lost some git-fu @ivanov ? you are doing weird stuff when you rebase. |
|
yeah, hang on, working on it |
- The `call` methods for nbconvert transformers has been renamed to `transform`. - The `call` methods of nbconvert post-processsors have been renamed to `postprocess`. pinging @Carreau and @jdrefer, who probably knows who'd be affected by this change... maybe @damianavila, @jakevdp
|
alright, how's that @Carreau |
|
Better, travis still in progress. |
|
are you not hip enough for hipchat, or too hip? |
|
This is starting to looks good. You have my green light to break master. When you'll push the green button, my EuroSciPy talk will no longer be compatible with master :-) |
|
Matthias Bussonnier, on 2013-08-16 14:22, wrote:
is that because your talk is more than meets the eye? |
|
Actually it's a 90min tutorial. I hope to have enought time to use nbconvert from CL and as python lib. Potentially defining custom transformer. Anyway I guess most people will run 1.0 from conda (I hope, as it is still not default in their repo), so I'll run 1.x too. |
|
Working with nbconvert APIs in the tutorial is going to be tricky, as they are guaranteed to change (as illustrated by the fact that it will already be broken on master by the time of the tutorial. We will have to tread lightly, and not emphasize APIs. |
|
Honestly, I don't know if I would even cover the nbconvert APIs right now On Fri, Aug 16, 2013 at 6:16 PM, Min RK notifications@github.com wrote:
Brian E. Granger |
Not in depth, but as soon as you speak architecture you use the name "Transfomer" All theses are touched by this PR. I try to avoid speaking of transformer/preprocesors as much as possible |
|
merging in a couple of hours unless I hear back otherwise |
rename call methods to transform and postprocess
rename call methods to transform and postprocess
The
callmethods for nbconvert transformers has been renamed totransform.The
callmethods of nbconvert post-processsors have been renamed topostprocess.pinging @Carreau and @jdrefer, who probably knows who'd be affected by
this change... maybe @damianavila, @jakevdp