| From: | "Robert Haas" <robertmhaas(at)gmail(dot)com> |
|---|---|
| To: | "Gregory Stark" <stark(at)enterprisedb(dot)com> |
| Cc: | "Tom Lane" <tgl(at)sss(dot)pgh(dot)pa(dot)us>, "Bruce Momjian" <bruce(at)momjian(dot)us>, "KaiGai Kohei" <kaigai(at)ak(dot)jp(dot)nec(dot)com>, "Martijn van Oosterhout" <kleptog(at)svana(dot)org>, "Alvaro Herrera" <alvherre(at)commandprompt(dot)com>, pgsql-hackers(at)postgresql(dot)org, "KaiGai Kohei" <kaigai(at)kaigai(dot)gr(dot)jp>, simon(at)2ndquadrant(dot)com |
| Subject: | Re: Updates of SE-PostgreSQL 8.4devel patches (r1403) |
| Date: | 2009-01-14 18:45:11 |
| Message-ID: | 603c8f070901141045h604ade55k14b4e446dfb00663@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
> It's not in C89 but look up "alloca".
I know about alloca...
> We don't use it anywhere in postgres currently so it's kind of unlikely we
> would start now.
:-(
>> Obviously this is a bad plan if x can be a big number because you
>> might crash your stack, but suppose we know that's not an issue? It
>> seems a shame to have to do palloc/pfree in a situation like this.
>
> palloc really isn't that expensive, unless you're allocating tons of tiny
> objects or you're in a tight loop it's not worth worrying about.
Yeah... but...
It really depends on what you compare it to. It's cheap compared to
99% of the functions in the code base - perhaps so. But it's darn
expensive compared to moving the stack pointer. I have seen profiles
for PostgreSQL and other systems where memory management is a sizable
percentage of the CPU time, so it is not silly to worry about
economizing.
...Robert
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Martijn van Oosterhout | 2009-01-14 19:14:15 | Re: Updates of SE-PostgreSQL 8.4devel patches (r1403) |
| Previous Message | Stephen Frost | 2009-01-14 18:43:06 | Re: New patch for Column-level privileges |