Re: [HACKERS] buildfarm failure after ICC configure change
Jeremy Drake [EMAIL PROTECTED] writes: On Wed, 12 Sep 2007, Tom Lane wrote: Argh! Can someone quote chapter and verse from the ICC manual about this? I was just following what Sergey said was the approved spelling of the switch ... Here are the docs for the two options. So why have they got two switches for what seems to be the same thing? I think I'll just revert the configure change. We know the gcc-compatible spelling works ... regards, tom lane ---(end of broadcast)--- TIP 2: Don't 'kill -9' the postmaster
[HACKERS] buildfarm failure after ICC configure change
I just saw that my buildfarm member (running ICC 9.0 on linux) failed after the latest change to configure http://www.pgbuildfarm.org/cgi-bin/show_log.pl?nm=mongoosedt=2007-09-11%2020:45:01 I was the one who sent in the first patch to configure to add the check for ICC, and as I recall at the time, the docs said -fno-strict-aliasing was the default (which is why I said not sure if this is needed in the comment). I don't know what -fno-alias does, but I think it means that the program does not do aliasing at all. The docs say You must specify -fno-alias if you do not want aliasing to be assumed in the program The true option for -fno-strict-aliasing is -ansi-alias- disable use of ANSI aliasing rules in optimizations. But this is the default... -- Help me, I'm a prisoner in a Fortune cookie file! ---(end of broadcast)--- TIP 3: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faq
Re: [HACKERS] buildfarm failure after ICC configure change
Jeremy Drake [EMAIL PROTECTED] writes: I just saw that my buildfarm member (running ICC 9.0 on linux) failed after the latest change to configure Argh! Can someone quote chapter and verse from the ICC manual about this? I was just following what Sergey said was the approved spelling of the switch ... regards, tom lane ---(end of broadcast)--- TIP 4: Have you searched our list archives? http://archives.postgresql.org
Re: [HACKERS] buildfarm failure after ICC configure change
On Wed, 12 Sep 2007, Tom Lane wrote: Jeremy Drake [EMAIL PROTECTED] writes: I just saw that my buildfarm member (running ICC 9.0 on linux) failed after the latest change to configure Argh! Can someone quote chapter and verse from the ICC manual about this? I was just following what Sergey said was the approved spelling of the switch ... The docs are not particularly clear about these options, at least not in 9.0 (which is the version I have). I figured -ansi-alias- was what -fno-strict-aliasing meant, that the gcc people decided to follow the ISO C standard strictly unless you gave -fno-strict-aliasing, which could result in additional optimization. This seems to match what the docs for -ansi-alias, but with a different default. The -fno-alias seems to go entirely the other way, saying you are not going to be doing any aliasing, standard permitted or otherwise, so it can optimize better accordingly. Here are the docs for the two options. == ansi-alias, Qansi-alias Enable use of ANSI aliasing rules in optimizations. Syntax Linux: -ansi-alias -ansi-alias- Windows: /Qansi-alias /Qansi-alias- Default -ansi-alias-Disable use of ANSI aliasing rules in optimizations. This option tells the compiler to assume that the program adheres to ISO C Standard aliasability rules. If your program adheres to these rules, then this option allows the compiler to optimize more aggressively. If it doesn't adhere to these rules, then it can cause the compiler to generate incorrect code. == falias Specifies that aliasing should be assumed in the program. Syntax Linux: -falias -fno-alias Windows: None Default -falias Aliasing is assumed in the program. This option specifies that aliasing should be assumed in the program. You must specify -fno-alias if you do not want aliasing to be assumed in the program. Alternate Options Linux: None Windows: /Oa -- Remember: Silly is a state of Mind, Stupid is a way of Life. -- Dave Butler ---(end of broadcast)--- TIP 5: don't forget to increase your free space map settings
