atstype_boxed(postiats_tyrec_123) atslab__54321

[Barry Schwartz]

I should have saved an example, but didn’t, because the code that
produced this (despite passing type checking) was a buggy
mess. Anyway, what causes me to get C code like

atstype_boxed(postiats_tyrec_123) atslab__54321 ;

that then fails in the C compiler? Apparently because the compiler
does not expect to see an argument to atstype_boxed.

[gmhwxi]

I so wish you had kept the source  for this :)

I took a look at the source code for patsopt but could not
figure out how to generate such output. I can generate something
like

atstype_tybox(...)

but that is really a different issue.

[Barry Schwartz]

> I so wish you had kept the source for this :)

D’oh!

The code has structures like @(key, value, list (@(key, value), n))
where key and value are t@ypes (separate chains for hashing), and I
think the bad code was generated for the list field of the outer tuple
when I was trying to do opt_some and/or Option_vt with one of these.

[Barry Schwartz]

gmhwxi <gmh...@gmail.com> skribis:

> I so wish you had kept the source for this :)

In the future I think I will check such stuff in on branches.

[Hongwei Xi]

Thanks!

I tried a few more examples following what you described, but could still not trigger the problem.

When did you noticed the problem? I remember fixing a problem involving atstype_boxed/atstype_unboxed

recently (within last two weeks).

--
You received this message because you are subscribed to the Google Groups "ats-lang-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-lang-user...@googlegroups.com.
To post to this group, send email to ats-lan...@googlegroups.com.
Visit this group at http://groups.google.com/group/ats-lang-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/ats-lang-users/20151004160853.GB8095%40crud.

[Barry Schwartz]

Hongwei Xi <gmh...@gmail.com> skribis:

> I tried a few more examples following what you described, but could still
> not trigger the problem.
>
> When did you noticed the problem? I remember fixing a problem involving
> atstype_boxed/atstype_unboxed
> recently (within last two weeks).

I was using 0.2.3 and didn’t try the Git sources.

[Barry Schwartz]

Hongwei Xi <gmh...@gmail.com> skribis:

> I tried a few more examples following what you described, but could still
> not trigger the problem.

Truly sorry I did not realize it wasn’t my fault. :-/