Closed susan-garry closed 9 months ago
Excellent; all sounds great!
3. Add automated testing via
runt
I see the new directory of tests, which seems great, but I don't set see a configuration file for Runt (or Turnt). Did this perhaps just not get checked in?
5. Rename the pollen extension to
.pollen
from.pol
..pol
is much shorter, but apparently is already a file extension for "group policy settings." I'd be curious if anyone has feeling about.pol
versus.pollen
.
Sounds right to me! Less ambiguity => more better.
I see the new directory of tests, which seems great, but I don't set see a configuration file for Runt (or Turnt). Did this perhaps just not get checked in?
Ah, it looks like runt.toml
was already merged into the main branch as part of #136 . I'll go ahead and update this PR to reflect that
The short change list:
flip.pollen
.{ e with f1:e1, f2:e2, ... }
)runt
~ All automated tests pass after #140test
folders totests
, for the sake of consistency with thecalyx
directory (I don't have a preference between one or the other, I just want to be consistent so that I don't have to think about which one I need depending on which directory I'm in).pollen
from.pol
..pol
is much shorter, but apparently is already a file extension for "group policy settings." I'd be curious if anyone has feeling about.pol
versus.pollen
.The longer explanation:
The majority of updates to the pollen examples consisted of
outset
withparset
,(t1*t2)
rather than(t1, t2)
.x += y
withx = x + y
(we do not yet support this syntactic sugar)in_g.paths
,in_g.steps
, andin_g.segments
with simplypaths
,steps
, andsegments
, though perhaps we would only want one special variable as opposed to 5 (includinghandles
andedges
).Additionally, I fixed a logical bug in
flip.pollen
related to ourparset
semantics - previously, it had been attempting to update a step's rank by iterating throughrev(path.steps)
and then emitting the step with it's orientation reversed without updating its index. However, as per our relational semantics, the order in which we emit steps does not matter, and we must instead specify its new index onpath
when we emit it, like so:{ step with idx: max_step_idx - step.idx }
.What this pull request does not do is completely reconcile our pollen tests with the parser's semantics -
paths.pollen
refers to the steps in the path viapath.steps[i]
, but the parser does not currently support array access. See #140