[m-rev.] for post-commit review: fix doc in term_io.m

Zoltan Somogyi zoltan.somogyi at runbox.com
Fri Apr 7 00:56:35 AEST 2023


For review by anyone.

The code of term_io.m was written before we had stream.string_writer,
so for some operations, it implements them twice (one version doing I/O,
one constructing a string), while for others, the string version is missing.
Would anyone object if I rewrote its guts using stream ops, the same way
that parse_tree_out_*.m now has write_X, X_to_string, and format_X
for various X? It would be slower than the current code at lower-than-default
optimization levels, but two typespec pragmas for all affected predicates,
enabled in Mercury.options, should cure this.

I also think we need a better naming scheme that allows people
to look at a pred or func name and know whether that op (a) quotes
the entity in question AND whether it (b) puts quotes around it or not.
At the moment, most names answer just one of those questions,
not both. But that is for a different diff.

Zoltan.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Log.tio
Type: application/octet-stream
Size: 350 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20230407/e39a8bd2/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: DIFF.tio
Type: application/octet-stream
Size: 7589 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20230407/e39a8bd2/attachment-0001.obj>


More information about the reviews mailing list