Closed ChrisJefferson closed 8 months ago
One reason this particularly came up is with the tutorials, selecting "re-run all cells" in them breaks, but "re-run all cells" doesn't reset the current model.
I'm going to make a contraversal suggestion -- switching to %%conjure_continue
(or %%conjure_extend
?), to continue building up a model, and switching %%conjure
to mean "start a fresh model". The reason to suggest that is at the moment using %%conjure
causes weird issues whenever you "rerun all cells", due to the lack of resetting
Thanks for bringing this up Chris, this has been bugging me for a while.
I think it would be ideal if we had a lighter weight solution in terms of new magics. How does this sound:
%%conjure --optional-cli-options
always creates a new model
%%conjure + --optional-cli-options
always adds to the existing model.
Is it too subtle?
if %%conjure+
is possible, that could be better as well...
%%conjure+
is nice if it works.
Could add %%conjure+
now, leave %%conjure
as is for now (while various models get changed here, and in other places), then change how %%conjure
after changes have permeated around?
Also, the extension is versioned & has multiple releases, so I am less worried about making breaking changes. I guess we could bump the version to 0.1.0 with this as well. I shall investigate, but I have a feeling %%conjure+
won't be possible. It would work with a space of course.
I see it is possible to use %%conjure+
through creating alias. There should be a way to do it programmatically.
oooh, that's nice!
I think I was able to implement conjure+
. Should %%conjure
reset the state of %%conjure+
?
I think that's the idea, yes. conjure
starts a new model, conjure+
appends to the existing model. The rest of the logic is identical. Right @ChrisJefferson? Thanks @ogabek96!
also: I guess we can remove conjure_clear
now?
I think this is now fixed via conjure+
At the moment, to make a new model, we seem to need a cell with just
%conjure_clear
, then a new%%conjure
cell.Maybe
%%conjure_new
?