Zilin Chen
Zilin Chen
91101fc doesn't match the Isabelle spec. In the spec when it checks the kind of `TCon tn ts s`, it looks at both `ts` and the sigil `s`, whereas in...
`#T` can either mean a stack object which contains pointers to the heap, or a purely stack allocated object. From the type we can't tell them apart. But their linearity...
More specifically, the stack of error contexts seems mostly missing at the moment.
Another specific case, when taking a field which doesn't belong to a record, it doesn't say so (or not explicit about it).
We don't have to fix it right now. I just wanted to know how things worked in order to make `make test-libgum` pass.
**Comment by [ahixon](https://github.csiro.au/ahixon)** _Thu Apr 9 17:48:59 2015_ ---- Unlikely to affect iozone benchmarks; more likely to change Filebench numbers.
**Comment by [ahixon](https://github.csiro.au/ahixon)** _Fri Apr 10 13:42:33 2015_ ---- Orlov only affects directory inode allocations.
**Comment by [ahixon](https://github.csiro.au/ahixon)** _Wed Apr 15 17:18:15 2015_ ---- Stated on this. Lives in ext2-orlov branch.
also see `cogent/lib/gum/common/vector.cogent` on `bilby-buffer` branch. `modref` ~ `focus_vector`.
Suggest using `upcast` maybe? The constraint `U16 :< U32` is actually satisfiable.