diff options
| author | Kartik K. Agaram <vc@akkartik.com> | 2015-05-25 20:33:37 -0700 |
|---|---|---|
| committer | Kartik K. Agaram <vc@akkartik.com> | 2015-05-25 21:36:46 -0700 |
| commit | b2757892d553352feb59d70b1e7241ccdafa6905 (patch) | |
| tree | 5a7597e46daf2db17d2fc5ade66aef3f07c261ce /045closure_name.cc | |
| parent | 64ac91c564a20e2ebf19ad211e987fbfc7fb2719 (diff) | |
| download | mu-b2757892d553352feb59d70b1e7241ccdafa6905.tar.gz | |
1458
While pushing out color support in fake screens I realized I've been complecting the special-case of a special-case to transform literal-string arguments for 'new'. As a result I hadn't been catching bad habits like giving its arg the wrong type. Now we have cleaner separation of the two variants of 'new', a few more checks, and better error messages when we mis-call it. Aside: I've added a third goto target. Sliding into spaghetti? Keep an eye on it. This goto might become a common pattern: a layer hooking into a previous one to prevent it from happening. In this case new on literal-strings prevents the transform for new from triggering.
Diffstat (limited to '045closure_name.cc')
| -rw-r--r-- | 045closure_name.cc | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/045closure_name.cc b/045closure_name.cc index 237eee3c..66696fc6 100644 --- a/045closure_name.cc +++ b/045closure_name.cc @@ -21,7 +21,7 @@ recipe init-counter [ ] recipe increment-counter [ - default-space:address:array:location <- new space:literal, 30:literal + default-space:address:array:location <- new location:type, 30:literal 0:address:array:location/names:init-counter <- next-ingredient # outer space must be created by 'init-counter' above y:number/space:1 <- add y:number/space:1, 1:literal # increment y:number <- copy 234:literal # dummy |