Lasse Fister
Lasse Fister
Pattern matching on the type of the check arguments comes to my mind. Could be done by some kind of generic switching meta-check that wraps the actual implementations. Maybe a...
Well, at the moment we have clearly defined memory leaks, they are called caches ;-) > We should be freeing up memory as we go see also: > There are...
(transfered from https://github.com/googlefonts/fontbakery/issues/1609#issuecomment-335632139) > It seems to me that we should be freeing up memory as we go. Yeah, that's clearly due to the caching of the `@condition`s I guess...
> We should also try to reduce excessive caching that makes the program die or become sluggish. Sure we need to do something; but it's not easy. E.g. as the...
> Eg, the runner runner can pass an argument to moderate or disable the runner's caching? Yeah, I'd go with an CLI argument first. We could try to infer a...
The collection-wide/super-family discussion is #1609 please talk about caching here and about the other thing there ;-) (updated: issue reference above)
> So, when the root input to the runner is the collection-wide set of ttFonts, then this not intended; Right, but "collection-wide set of ttFonts" was not intended in the...
OK, but it was never implemented. I'm talking about the implementation and actual changes to it.
2.5 seems missing, the rest is, if 2.3 is ignored (which we chose to I think)
Maybe we should make CheckRunner into a Family-CheckRunner (just a name change really, or even better don't change the name, just use the concept!), and add a CollectionCheckRunner, that handles...