You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The OFP frontend is getting increasingly hard to maintain. A recent brush with some of its inconsistencies (mostly due to skew in the generated AST node scoping due to upstream bugs) has highlighted that in particular better handling of functions/procedures will become very painful if we are to maintain this.
I've checked and the most recent release (0.6.1) is not compatible with our frontend implementation, but it does still have many of the scoping quirks that we tend to bend over backwards to circumvent. As we have little use for it that we cannot deal with via other frontends, I propose we now fully deprecate and/or entirely remove the OFP frontend. This should improve test throughput and remove some of the special-casing, ideally before we continue to align procedures as proper IR nodes.
I'm happy to implement and test this, but input and feedback are very welcome (@reuterbal ).
The text was updated successfully, but these errors were encountered:
I agree. OFP is dead slow, which, among all the other issues, would make it undesirable for any productive use anyway. There is value in maintaining more than one frontend but arguably we should rather spend time on exploring other alternatives than fixing up OFP.
The OFP frontend is getting increasingly hard to maintain. A recent brush with some of its inconsistencies (mostly due to skew in the generated AST node scoping due to upstream bugs) has highlighted that in particular better handling of functions/procedures will become very painful if we are to maintain this.
I've checked and the most recent release (0.6.1) is not compatible with our frontend implementation, but it does still have many of the scoping quirks that we tend to bend over backwards to circumvent. As we have little use for it that we cannot deal with via other frontends, I propose we now fully deprecate and/or entirely remove the OFP frontend. This should improve test throughput and remove some of the special-casing, ideally before we continue to align procedures as proper IR nodes.
I'm happy to implement and test this, but input and feedback are very welcome (@reuterbal ).
The text was updated successfully, but these errors were encountered: