Closed moodymudskipper closed 3 years ago
current opinion with a fresh look after a few months, agree mostly with above :
This old thread has been automatically locked. If you think you have found something related to this, please open a new issue and link to this old issue if necessary.
confusing ellipsis
as arguments pile up, having
...
forwarded to nomnoml doesn't look good, and it's not that flexible either.An argument
engine.opts
expecting a list of parameters, which could default togetOptions("flow.engine.opts")
so these could be set globally too.If the list has names "plantuml" and "nomnoml" we'll take the right item depending on the
engine
param (so user can set global options that work for all engine). If not we'll feed the list as is to the relevant engine function.clunky range argument
We might remove it altogether, and add later if needed. It's only useful for huge functions for which the diagram engines are the limitation. In practice that's very few.
width and height
They're forwarded to hmlwidget but it's not clear how they work with flow.
probably better either combine them in a list of args to forward to htmlwidget, orjust remove altogether
sub_fun_id
now that we allow names and not only ids, the name is not great, if we modify the api maybe take the opportunity to change this argument name too ?
sub_function
?sub_fun
?nested_fun
?Another good consequence of these changes is that it would harmonize the api for nomnoml and plantuml as our list of args wouldn't be cluttered by args that apply only yo nomnoml. They'd be in the end as
nomnoml.args
andhtmlwidgets.args
, and maybestyler.args
(but maybe better not to add too much flexibility upfront).I think we might unexport flow_data and build_nomnoml data too. The users don't really care at this point and it's confusing re plantuml vs nomnoml.