* Number 'view.patch' output
* Add numbering to patches
* Change transcript to use numbers rather than hashes
* Update transcripts
* use short-hash as lhs of patch outputs in numberedArgs
A doc containing an embedded term link in a doc like
`{{a doc linking to {abilityFoo}}}` would previously eagerly lex
`ability` or `type` as a keyword, tricking the parser into thinking
we're in an embedded type link, and then complain about not finding
`Foo`. This fixes the lexer to treat such cases as term links instead.
Adds a regression test transcript for making sure that `type` and
`ability` are lexed properly in type and term links in docs.
Annotations still need to be stripped off for evaluation, but we can
keep them around in some of the code stored for decompilation. It
requires some delicate changes to the lambda lifting process, though.
this is a partial fix for #2834 more will have to come. Now there is a
cert there valid for 10 years and we load it and the private key with
now problem, the call to decode the cert now is checking for
exceptions and is properly handling the case when we successfully
decode 0 certs.
Now for some reason the last section of the transcript where we try to
do handshaking, we are Segfaulting on the one of the sides of the
handshake. I haven't figured out why yet.
It seems that optimizations cause problems together with the combination
of coercing and unsafe IO involved in reference equality. I'm not
exactly sure what goes wrong, but it causes segfaults. Defining a
noinline function for the equalities (instead of just `(==)@Reference`)
seems to fix the problem.
I got the same problem to happen for term links, so both of those use a
noinline function now. Other foreign values don't seem to be affected,
which I assume is due to them having pure primitive functions for
equality.
- This commit shuffles the order of how interactive evaluataion happens
to attempt to get decompiler output that is closer to the original
source.
- It only does floating before remembering the combinators, so output
shouldn't have compiled patterns and the like anymore.
- I was hoping that this might preserve some type annotations in
embedded docs, but it appears to not be the case.