| commit | 455e207d461735fd6da1cbdfdcd9b23a32da2d4a | |
| author | Douglas Katzman <dougk@google.com> | |
| Thu, 4 May 2017 15:59:14 +0000 (4 11:59 -0400) | ||
| committer | Douglas Katzman <dougk@google.com> | |
| Thu, 4 May 2017 17:08:36 +0000 (4 13:08 -0400) | ||
| tree | 1b4b32e0995fa07342c6a6cb6ae0d9761243acf7 | treesnapshot (tar.gz zip) |
| parent | 54377a283616be8c4115dfd278240695433a78ad | commitdiff |
Better sharing of symbol print-names.
Do more thorough collapsing because:
"X3J13 voted in March 1989 to specify that it is an error
to modify a string being used as the print name of a symbol."
and remove logic from Genesis that did the same thing.
Also prefer to use base strings in the common pattern of
(INTERN (FORMAT NIL "~A-~A")). While it would be nicer to have INTERN
always pick the minimal space representation no matter what,
it seems like some people think that (INTERN "FOO") - though permitted
to copy "FOO" - must copy to the same type string, which is unspecified
as far as I can tell. That said, it's unarguable that (FORMAT NIL ...)
produces any particular type of string.
And this saved us >7MB in our application, so ... there you go.
Do more thorough collapsing because:
"X3J13 voted in March 1989 to specify that it is an error
to modify a string being used as the print name of a symbol."
and remove logic from Genesis that did the same thing.
Also prefer to use base strings in the common pattern of
(INTERN (FORMAT NIL "~A-~A")). While it would be nicer to have INTERN
always pick the minimal space representation no matter what,
it seems like some people think that (INTERN "FOO") - though permitted
to copy "FOO" - must copy to the same type string, which is unspecified
as far as I can tell. That said, it's unarguable that (FORMAT NIL ...)
produces any particular type of string.
And this saved us >7MB in our application, so ... there you go.