Re: iso-8601 sub-seconds problem w/patch
"Hoehle, Joerg-Cyril" <[email protected]>
| Newsgroups | gmane.lisp.clsql.devel |
|---|---|
| Message-ID | <5F9130612D07074EB0A0CE7E69FD7A0303BE82BB@S4DE8PSAAGS.blf.telekom.de> |
Hi, > (parse-integer (subseq string 5 7)) PARSE-INTEGER features :start and :end keywords that make almost all of these SUBSEQ food for the garbage collector strictly superfluous. Funnily, pg.lisp (not clsql) also does not make use of these really handy keywords. People may argue that given generational collectors, things like avoiding garbage do not matter that much anymore (and wire speed etc. arguments). However even with gen. GC, the data to become garbage has to be allocated and *copied*. That is strictly superfluous waste of resources (which remembers me of that TCP/IP in ML paper where they found out that their speed was limited by all the copying of data). Avoiding copying in the first place is one key to performance. Regards, Jorg Hohle