Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] View of /papers/modulespaper/notes/issues.txt
ViewVC logotype

View of /papers/modulespaper/notes/issues.txt

Parent Directory Parent Directory | Revision Log Revision Log


Revision 3564 - (download) (annotate)
Thu Sep 30 13:33:05 2010 UTC (8 years, 9 months ago) by dbm
File size: 2783 byte(s)
initial import
Issues

0. have to take great care to avoid wasting text on explaining 
routine facts about figures and rules. Let the figures stand on their
own, and be very selective in choosing what points need further explanation.

1. Notation for tycon kinds: fig-typesystem.tex used "->" while other
figures and text used "=>".  I've changed it to "=>" in the figure.

2. It is conceivable to drop polymorphic types in the core language,
since it is a fairly secondary issue (Shao99 doesn't include them).
But it isn't clear that this would save a significant amount of space.

3. The well-kinding rules for syntactic and semantic types are
standard and routine, so I've deleted the corresponding figures and
limited the discussion to just noting that the languages have standard
well-kinding rules, which are omitted.

4. Same for discussion of normalization of type expressions.  This is
well-understood and routine (at least for our intended audience).

5. For things like the surface module language (fig:modlang), let the
reader study the figure. We don't need to waste a lot of text on 
reiterating what the syntax definitions say. Just briefly mention
highlights that may not be sufficiently obvious.

6. We will probably end up just summarizing the translation section,
and omitting all the corresponding figures. Budget at most one column
for this.  Will require careful summarizing.

7. Modified versions of dissertation figs are being moved to 
design/figs. One common modification is to remove the lower \hrule,
since the SIGPLAN figures already come with a lower \hrule.
Also renaming figure files to agree with the figure labels
(e.g. fig:modlang <-> figs/fig-modlang.tex --- was fig-microml.tex).

8. In syntax.tex introduced two abbreviations
     \Uloc for \Upsilon_{l}  (replacing \Upsilon^{lcl})
     \Uclo for \Upsilon_{c}  (replacing \Upsilon^{clo})
   Updated main.tex and relevant figures in design/fig, but
   there are probably figures in dissertation/figs that will
   need to be updated (and copied over to design/figs).

9. The rule for evaluating entity declarations for defined tycons
is incorrect (eq:entdectypedef in fig-entdec.tex).  It needs to
evaluate the relativized tycon expression against the entity
environment to get a nonrelativized (absolute) tycon. I didn't
put this rule in the paper -- it's relatively routine.
[There is a global invariant that all tycons in realizations,
or any entity environments are absolute (i.e. are not relativized).
Thus before you can add a tycon binding to an entenv its 
relativized definition has to be instantiated relative to the
context entenv.]

10. The three columns devoted to relativization (Sections
4.4 -- 4.6 are way to much.  This should be cut down to probably
less than a column -- maybe a half column.


root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0