Closed lucaswerkmeister closed 8 years ago
Currently, the formatter can format about 75% of FormattingOptions_generated.ceylon
.
The formatter now supports all language features used in FormattingOptions_generated.ceylon
. However, I’m going to postpone this in favour of even more language features so IDE integration can start ASAP.
Postponing after 1.0.0 because it needs ceylon.ast
, which is now my GSoC proposal.
Actually… this would make ceylon.formatter
depend on itself for building, making a bootstrapped build impossible (the “escape” would be an already-available version, e. g. on the Herd – but what if there are compatibility problems, binary or otherwise?). I’d say this is a great case of “never touch a running system”, i. e., our current source-gen (while a minor pain to write and, when needed, update), works well enough.
Closing this. Icky circular dependency, not necessary since codegen is already there.
With #17, we can use the formatter as a code writer for the source generation (
FormattingOptions_generated.ceylon
). This would be both a neat demonstration of its capabilities, and a big step towards dogfooding.I suspect I’ll also start a
ceylon.ast
module for this...