Re: [dylan] Looking for some prior art ... compiler testing, extending the compiler

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA384

Hi,

my needs are in two areas:
 - test cases for compiler (especially type inference and optimizer)
 - "improvements" of an unchanged library from a previous compiler run

The first is mostly straightforward - a query language for control and
data flow might be handy - but sometimes I'd also want to specify the
complete (apart from alpha-conversion) control and data flow graphs
for a given test case (basically a reader for DFM). Also, I'd want to
specify which optimizations to run for a test case, otherwise useful
information might disappear.

The second is basically a diff on some structured data - namely the
control flow graph. Here the diff must be pretty smart: the control
flow graph is a cyclic graph (read: loops), which can fortunately be
converted to a acyclic graph by removing the edge from continue to
loop entry.

But a diff for DAGs is hard. Especially since there are costs
connected to each control flow node (a generic function call (= 1
node) is more expensive than the inlined method (= 100 nodes,
containing 10 generic function calls).

I don't really see the advantage of using json or whatever format here
instead of DFM (saved to/restored from disk using DOOD). (Obviously
the text/ASCII representation of DFM is not complete or "good" -
that's a reason why I implemented a graph visualization (which is
useful for debugging, but not for batch runs)).


Cheers,

Hannes

On 08/02/2013 10:51, Bruce Mitchener wrote:
> - Test the optimizer and type inference in varying combinations. -
> Record the current state of code generation so that various
> versions of the compiler output can be diffed. - Implement new
> optimizations. - Implement new warnings. - Implement static
> checkers.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (Darwin)
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=tPio
-----END PGP SIGNATURE-----
_______________________________________________
hackers mailing list
[email protected]
https://lists.opendylan.org/mailman/listinfo/hackers