Re: [gentoo-user] Re: Is perl broken?
On Saturday 11 Apr 2015 00:08:23 Peter Humphrey wrote: Back to the original theme, I'd been experimenting with -j and -l make options, and I suspect that was my real problem. I finished up with -j -l20 on this i5 box, with startling results - 56 emerges in parallel for instance. I suspect that my problem stemmed from this. All now seems stable so far with -j12 and no -l specified. Satisfactory CPU utilisation and the all-important stability. So no, perl isn't broken :) On an early generation mobile i7 I have MAKEOPTS=-j5 -l12.8 with no noticeable adverse effects, although on recent mammoth emerges the responsiveness goes down when swap is used. I can't even recall how I arrived at these settings. -- Regards, Mick signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] Re: Is perl broken?
This is a re-send of a message I sent earlier today but which seems not to have appeared on the list - well, I have changed it a bit: On Tuesday 07 April 2015 23:19:18 I wrote: On Tuesday 07 April 2015 15:02:36 walt wrote: On 04/07/2015 02:48 PM, Peter Humphrey wrote: On Tuesday 07 April 2015 22:24:38 Peter Humphrey wrote: $ cat make.conf # I made a local copy and removed a lot of comments #CFLAGS=-O2 -march=core2 -pipe [1] ---8 [1] This bothers me. Various docs tell me to specify march=corei7, but this is an i5 CPU. Could this be my problem? Any reason you don't want to use march=native? Not that I can think of now. I'll try it - thanks, both of you. Countless CPU cycles later, I have now reinstalled my complete system with -march=native. It took several iterations. Meanwhile,I had another problem to keep me amused - KMail decided I'd deleted the folder into which it receives all inbound mail. I hadn't, of course, but suddenly my 13000 mails were gone - vanished. So I had to create a new user and import them all from the previous day's backup. Tedium - yawn... Still, all my filters have gone, and I'll have to define new ones as I need them. Oh well, I suppose it's about time I cleaned them out. Back to the original theme, I'd been experimenting with -j and -l make options, and I suspect that was my real problem. I finished up with -j -l20 on this i5 box, with startling results - 56 emerges in parallel for instance. I suspect that my problem stemmed from this. All now seems stable so far with -j12 and no -l specified. Satisfactory CPU utilisation and the all-important stability. So no, perl isn't broken :) -- Rgds Peter
[gentoo-user] Re: Is perl broken?
On 04/07/2015 02:48 PM, Peter Humphrey wrote: On Tuesday 07 April 2015 22:24:38 Peter Humphrey wrote: $ cat make.conf # I made a local copy and removed a lot of comments #CFLAGS=-O2 -march=core2 -pipe [1] ---8 [1] This bothers me. Various docs tell me to specify march=corei7, but this is an i5 CPU. Could this be my problem? Any reason you don't want to use march=native?
Re: [gentoo-user] Re: Is perl broken?
On Tuesday 07 April 2015 15:02:36 walt wrote: On 04/07/2015 02:48 PM, Peter Humphrey wrote: On Tuesday 07 April 2015 22:24:38 Peter Humphrey wrote: $ cat make.conf# I made a local copy and removed a lot of comments #CFLAGS=-O2 -march=core2 -pipe [1] ---8 [1] This bothers me. Various docs tell me to specify march=corei7, but this is an i5 CPU. Could this be my problem? Any reason you don't want to use march=native? Not that I can think of now. I'll try it - thanks, both of you. -- Rgds Peter.
[gentoo-user] Re: Is perl broken?
Andreas K. Huettel dilfri...@gentoo.org wrote: Moreover, I didn't check before the rebuild, but after the rebuild there is no 5.20.1 in @INC. Sure about this? I checked this, of course. But now I realize that the path is *added* to @INC (even to the perl -V output!) when I re-create it...
Re: [gentoo-user] Re: Is perl broken?
-BEGIN PGP SIGNED MESSAGE- Hash: SHA512 Am Montag, 6. April 2015, 13:29:25 schrieb Martin Vaeth: Andreas K. Huettel dilfri...@gentoo.org wrote: Moreover, I didn't check before the rebuild, but after the rebuild there is no 5.20.1 in @INC. Sure about this? I checked this, of course. But now I realize that the path is *added* to @INC (even to the perl -V output!) when I re-create it... :) That's the mysterious OLDVERSEN variable in line 11 of the ebuild... https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/dev-lang/perl/perl-5.20.2.ebuild?revision=1.7view=markup - -- Andreas K. Huettel Gentoo Linux developer dilfri...@gentoo.org http://www.akhuettel.de/ -BEGIN PGP SIGNATURE- Version: GnuPG v2.0 iQJ8BAEBCgBmBQJVIoHfXxSAAC4AKGlzc3Vlci1mcHJAbm90YXRpb25zLm9w ZW5wZ3AuZmlmdGhob3JzZW1hbi5uZXQ0RkJDMzI0NjNBOTIwMDY5MTQ2NkMzNDBF MTM4NkZEN0VGNEI1Nzc5AAoJEOE4b9fvS1d5xLAP/1KLDNuTEOhlpAEbSTj4Jcuc gRtHMX3bziPfFZMuesCS+4W6ECtua/ySf+MZfOgeTBOY981cfJsLEiftJM8Egvqw tVJthXfmqrRSy04cFZXpZ0U9oGY2qaQIQbmROBSvM5AKGkk3NRMSTPYhqQca2akz AWe49mkYcWE1vmHZfAYo5pDK8EvawU9/XJxvQvgaOh7IvxAjyclBf3DEWeB6UMN5 4j9CNHfJYf2Ts1vlb0p5sWcUuhshOnPQw6RY/YaEwzIE6F8cxp7Ja7B/6eY5Jr9D 2LMdXyhAg7GFE338OJf2kXWEoqZ2H7/1tBePp+plYBdEnLBm0c145M4R6KmE85EJ ZhCfK1/IVOshZuH8/uh1Hxwf5TDBeuKcLW+hRSbyC/uha2fx0WLGTvcNBHLDNt/h 2X757UGEKyn4Nj7Iq6OcJq57Bzw9b1w3h1AYiyKEbkzDb0iEptSiPyieZvbHUOBG KexKTmfgAoYgbrrprulxQpv9YO4pQefzFQMPUXktNaMU32ehibqFSMEWeRXAUjmj QxbJAXZWOcEO8HWKxZQ0sdbOR1pWPzer6lIsVbFIKJ+QPls/+DQKQgZH/YraBzNN iodZslB6ervXphrn3p4XnNfE6U7LumtuPy0GdQdlkQqpAVxgjT6XZWJaE/Kb8BGM 8Q0uP0CYsqQzCmQyetR4 =Qg6n -END PGP SIGNATURE-
[gentoo-user] Re: Is perl broken?
Andreas K. Huettel dilfri...@gentoo.org wrote: Minor updates (5.x.y - 5.x.y+1) do not need any rebuilds or reinstallations of modules. This is at most partially correct: At least, after the update, the install directories change; here from /usr/lib/perl5/{vendor_perl,}/5.20.1 to /usr/lib/perl5/{vendor_perl,}/5.20.2 So, at least, perl-cleaner wants to rebuild, and it is sane to do this (for various reasons: avoiding confusion with mixed directories, compitability with binary packages, omitting redundant directories). Moreover, I didn't check before the rebuild, but after the rebuild there is no 5.20.1 in @INC. (So it might be even the case that the rebuild is *necessary*). I suggest to either use the same 5.x directory for all 5.x versions, or to include 5.x.y into the subslot name to avoid the above mentioned minor inconsistencies. After all, the final aim is to use subslots instead of perl-cleaner, isn't it?
Re: [gentoo-user] Re: Is perl broken?
-BEGIN PGP SIGNED MESSAGE- Hash: SHA512 Am Sonntag, 5. April 2015, 21:53:35 schrieb Martin Vaeth: Moreover, I didn't check before the rebuild, but after the rebuild there is no 5.20.1 in @INC. (So it might be even the case that the rebuild is *necessary*). Sure about this? huettel@pinacolada ~/Gentoo/office/app-text/writerperfect $ perl -V Summary of my perl5 (revision 5 version 20 subversion 2) configuration: [...] Built under linux Compiled at Feb 14 2015 23:56:45 @INC: /etc/perl /usr/local/lib64/perl5/5.20.2/x86_64-linux-thread-multi /usr/local/lib64/perl5/5.20.2 /usr/lib64/perl5/vendor_perl/5.20.2/x86_64-linux-thread-multi /usr/lib64/perl5/vendor_perl/5.20.2 /usr/local/lib64/perl5 /usr/lib64/perl5/vendor_perl/5.20.1/x86_64-linux-thread-multi /usr/lib64/perl5/vendor_perl/5.20.1 /usr/lib64/perl5/vendor_perl /usr/lib64/perl5/5.20.2/x86_64-linux-thread-multi /usr/lib64/perl5/5.20.2 . - -- Andreas K. Huettel Gentoo Linux developer dilfri...@gentoo.org http://www.akhuettel.de/ -BEGIN PGP SIGNATURE- Version: GnuPG v2.0 iQJ8BAEBCgBmBQJVIZW8XxSAAC4AKGlzc3Vlci1mcHJAbm90YXRpb25zLm9w ZW5wZ3AuZmlmdGhob3JzZW1hbi5uZXQ0RkJDMzI0NjNBOTIwMDY5MTQ2NkMzNDBF MTM4NkZEN0VGNEI1Nzc5AAoJEOE4b9fvS1d5QasQAIovcx6dpeqqsraW5Uss2tUu udWdBAKfDtztMjRvzKdIQT1JO84g8oB3FhnmoWAgzeHUIWfsormJ/6EbIPEmb+eF Dct9daFa3wEQhUBEV4Wr3YVSnOl6LMZ1ZOPtlkAqgmGMEz73yvTMrwwNMjhSAy5u 7KXooNfM5pvagfCUyWkXY/uUcCC3FhT6RaLzddJcFL4dikTD6lrKLdizwclnrbNJ YK9ZRETwsJjhyfYCoZxq5MpLMJrlstZVV++RPBv94tRbUPGdeWiie34XYG+GVQ34 lQ7oc0xDFNL9kT672uEd6ZJi7U5icM7DlruTNXNoYT3bZeo9+yKxqsvbxWzsppVS ekjKVkCjfEGW+Swk8wQDWZCLvjm+9Pz/RPR33Dk+sI5q/Xj/3jrSieANMGlpEEkF titj8peF/Z8Rmd9EAmPwx1j2fssXPDZLwYlJq5lTwtyVl2/lkCpeMesbf0wFMeJd TTveTkS17PT+Dde9ok6cQ4Z2e4lc6DxuQLWw2paCrYqnwirlWyXv3OD8p7VDc7+Q sCkgl3OcYtcRXkDjnqBmfWZdi8khAXu78NNqDpNxh1d/LqKX6kV3rxVvt1oS1pcA ViMMBMQXJvBSfhh7dfYLtrLz5ZqHXqvBJ+vcG2BkPcGwbhPD8iH1VvI/Q/JyBRNb NFtPsQjMud43SYTOVU8z =TpDv -END PGP SIGNATURE-