Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On 11/03/2012 09:29 PM, Ben de Groot wrote: On 4 November 2012 08:28, Ian Stakenvicius a...@gentoo.org wrote: 03.11.12 09:04, Ben de Groot написав(ла): I think we need to come up with a better policy regarding elog messages, which would improve the signal to noise ratio. As of EAPI=4 (i think? maybe 3?) ebuild authors can use a simple test to see if $REPLACING_VERSIONS is empty or not to indicate whether the package is being newly installed or if it's being upgraded. Maybe, if package maintainers are so inclined, we can start using this feature to reduce the spam. (if we were really motivated we could even only provide relevant messages based on the version being replaced!) I think that would be a good idea. I would like to see repoman complain about elog messages without REPLACING_VERSIONS. I think has_version calls are also a good indicator that the ebuild is doing the right thing. Parsing bash can be somewhat error-prone for repoman, some I'd suggest a simple rule like Warn if elog is called and the ebuild does not contain any calls to the has_version function or references to the REPLACING_VERSIONS variable. -- Thanks, Zac
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On 31 October 2012 23:17, Chris Reffett creff...@gentoo.org wrote: Basically, I would rather the user get too many elog messages than not enough, since I feel that a lot of people skip over them anyway and so the only display once method makes it far too easy for important messages to get lost in the shuffle. Did you ever consider *why* so many users skip over elog messages? Don't you think that could be because most of these messages of useless to them, or they have already seem them many times? I think we need to come up with a better policy regarding elog messages, which would improve the signal to noise ratio. -- Cheers, Ben | yngwin Gentoo developer Gentoo Qt project lead, Gentoo Wiki admin
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
03.11.12 09:04, Ben de Groot написав(ла): On 31 October 2012 23:17, Chris Reffett creff...@gentoo.org wrote: Basically, I would rather the user get too many elog messages than not enough, since I feel that a lot of people skip over them anyway and so the only display once method makes it far too easy for important messages to get lost in the shuffle. Did you ever consider *why* so many users skip over elog messages? Don't you think that could be because most of these messages of useless to them, or they have already seem them many times? I think we need to come up with a better policy regarding elog messages, which would improve the signal to noise ratio. Yes, it would be really great if messages that are not relevant for upgrading of a package will be shown only for the first install. For example, I've just finished upgrading of Gentoo on my notebook and I was spammed with heaps of messages. They are for 10 packages. From them messages for net-misc/dhcpcd, net-misc/tor, sci-physics/root, mail-client/thunderbird, www-client/firefox and sci-libs/scipy are typical and not relevant for update. It means for 6 from 10 packages (Only message for dev-libs/boost is quite original with its red color )) ). It is of course not very important issue, but it is quite annoying. signature.asc Description: OpenPGP digital signature
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
-BEGIN PGP SIGNED MESSAGE- Hash: SHA256 On 03/11/12 08:19 PM, Jauhien Piatlicki wrote: 03.11.12 09:04, Ben de Groot написав(ла): On 31 October 2012 23:17, Chris Reffett creff...@gentoo.org wrote: Basically, I would rather the user get too many elog messages than not enough, since I feel that a lot of people skip over them anyway and so the only display once method makes it far too easy for important messages to get lost in the shuffle. Did you ever consider *why* so many users skip over elog messages? Don't you think that could be because most of these messages of useless to them, or they have already seem them many times? I think we need to come up with a better policy regarding elog messages, which would improve the signal to noise ratio. Yes, it would be really great if messages that are not relevant for upgrading of a package will be shown only for the first install. For example, I've just finished upgrading of Gentoo on my notebook and I was spammed with heaps of messages. They are for 10 packages. From them messages for net-misc/dhcpcd, net-misc/tor, sci-physics/root, mail-client/thunderbird, www-client/firefox and sci-libs/scipy are typical and not relevant for update. It means for 6 from 10 packages (Only message for dev-libs/boost is quite original with its red color )) ). It is of course not very important issue, but it is quite annoying. As of EAPI=4 (i think? maybe 3?) ebuild authors can use a simple test to see if $REPLACING_VERSIONS is empty or not to indicate whether the package is being newly installed or if it's being upgraded. Maybe, if package maintainers are so inclined, we can start using this feature to reduce the spam. (if we were really motivated we could even only provide relevant messages based on the version being replaced!) -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.19 (GNU/Linux) iF4EAREIAAYFAlCVtrUACgkQ2ugaI38ACPDOAgD+ORyGkW/Rqx4BoRWl6+172Edu MKeVblQYLyHIkL9Y0zEA+wYDMgQblhfcHKEMW9cgLsFDkWt0dF5KiUVd1fV1SIT9 =TWem -END PGP SIGNATURE-
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On 4 November 2012 08:28, Ian Stakenvicius a...@gentoo.org wrote: 03.11.12 09:04, Ben de Groot написав(ла): I think we need to come up with a better policy regarding elog messages, which would improve the signal to noise ratio. As of EAPI=4 (i think? maybe 3?) ebuild authors can use a simple test to see if $REPLACING_VERSIONS is empty or not to indicate whether the package is being newly installed or if it's being upgraded. Maybe, if package maintainers are so inclined, we can start using this feature to reduce the spam. (if we were really motivated we could even only provide relevant messages based on the version being replaced!) I think that would be a good idea. I would like to see repoman complain about elog messages without REPLACING_VERSIONS. -- Cheers, Ben | yngwin Gentoo developer Gentoo Qt project lead, Gentoo Wiki admin
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On Wed, Oct 31, 2012 at 5:08 PM, Jeroen Roovers j...@gentoo.org wrote: Though if you don't know these kinds of basics, I'm not sure you should be doing *any* (semi- or not) automated bug filing. What if you don't have the privilege to assign bugs, but are willing to do the work of filing the bugs? Yeah, that kind of sucks. Perhaps we should extend the privilege a little more often? Cheers, Dirkjan
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 11/01/2012 03:59 AM, Dirkjan Ochtman wrote: On Wed, Oct 31, 2012 at 5:08 PM, Jeroen Roovers j...@gentoo.org wrote: Though if you don't know these kinds of basics, I'm not sure you should be doing *any* (semi- or not) automated bug filing. What if you don't have the privilege to assign bugs, but are willing to do the work of filing the bugs? Yeah, that kind of sucks. Perhaps we should extend the privilege a little more often? I agree 100%, if anyone is willing to file proper and correct bugs in an automated way then we need to extend to them the ability to properly assign them to prevent our bug-wranglers from being crushed. I believe that allowing someone to assign a bug is very appropriate in this case. - -Zero_Chaos Cheers, Dirkjan -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.19 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://www.enigmail.net/ iQIcBAEBAgAGBQJQkmnDAAoJEKXdFCfdEflKcgsP/j5V0RhZcrJ2qGok8hlZtAdy +0D1uqbp82XUud6esUgNkCMueFrKHVKURtZwIQlUiL4Y5MexWsPk5/+V0iH+XMGl H0Sc5KlDTnjB9Fi67/026hm2FvAzy2nyeoXazqdc8jnhG4/adMp5UdRcjPtH+nLm 3KShd2BIava1lSZRR3EEOlo2qPFtYloxy7KsGWN+KwC+03DLEyCyeJbN06yxu5B9 z7eo53nSvig6aZKQTwd+6n9MKjHUDkHK+Hv3y7q2S4IR1+3gC7kvrTqHE+LheCVO sZ+BQzGwNUuAeVvmdlJSXLMuQRmR5SoWv7CqRccQcLFX/AX8hhyvQaQidAP3YzG/ gZqx7dc0GvWJ0E1bHQeN0/6YhfvnR83TEaugfilp+BzLfvqHLY9xCtVNiGzYQBqS CjGTeJqH2lPhpg7+/cyKU6l17tAN0TjRd3y9K9F52QJskwth0Rs3qxBT/Ol/zGqC rYDyOPGxu+/cgYFGJn796OZEedh9/kbk3GnvOdAO9yUpU/1JuiCnZ+1l95DJfrzi eFSegwB3DTy7Pr0GNgp3R1vFVlwLL7AUIfHVbdrY17xqwx/Om1m1WdeqyZIWymp3 Lc+riJQBNbyJ0NKX/7joK8JOyPK7SzHBSzWgOl0PYxw/RzdKVv4BSWOAVVwb4Iof JTozrERY+GUb7Rm1IGKt =4Fqk -END PGP SIGNATURE-
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On Thu, Nov 1, 2012 at 7:59 AM, Dirkjan Ochtman d...@gentoo.org wrote: On Wed, Oct 31, 2012 at 5:08 PM, Jeroen Roovers j...@gentoo.org wrote: Though if you don't know these kinds of basics, I'm not sure you should be doing *any* (semi- or not) automated bug filing. What if you don't have the privilege to assign bugs, but are willing to do the work of filing the bugs? Yeah, that kind of sucks. Perhaps we should extend the privilege a little more often? Cheers, Dirkjan Extend it? More often? It is not a matter of frequency. If a user requests bugzilla rights, and a developer trusts him, he can grant him access. -- Regards, Markos Chandras / Gentoo Linux Developer / Key ID: B4AFF2C2
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On Thu, 1 Nov 2012 08:59:09 +0100 Dirkjan Ochtman d...@gentoo.org wrote: What if you don't have the privilege to assign bugs, but are willing to do the work of filing the bugs? Yeah, that kind of sucks. Perhaps we should extend the privilege a little more often? In this case no prior history helped me and I foresaw only a few bug reports arising, not several dozens. And anyway, assigning them took only half an hour. jer
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
Simply look at the metadata.xml files which can be found in each package directory. Though if you don't know these kinds of basics, I'm not sure you should be doing *any* (semi- or not) automated bug filing. Cheers, Dirkjan On Wed, Oct 31, 2012 at 1:34 PM, Marco Poletti poletti.ma...@gmail.com wrote: On mer 31 ott 2012 13:31:53 CET, Dirkjan Ochtman wrote: That's rather unsurprising... If you're going to file bugs in a semi-automated manner, might as well try to assign to the correct maintainer? Can you give me some help on how to do that? How can I find the maintainer of a package? Marco
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
2012/10/31 Dirkjan Ochtman d...@gentoo.org: That's rather unsurprising... If you're going to file bugs in a semi-automated manner, might as well try to assign to the correct maintainer? Yep he should've assign them, but anyway the annoying elog messages are an issue. And quite few packages suffer from it :-) Tom
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On Wed, Oct 31, 2012 at 8:31 AM, Dirkjan Ochtman d...@gentoo.org wrote: If you're going to file bugs in a semi-automated manner, might as well try to assign to the correct maintainer? How about a policy - no automated bugs may be logged to the bug wranglers without their prior approval/review. I wouldn't think about running a script against somebody's bugzilla/repository/forum/whatever without some kind of prior communication with the administrators. This usually is considered abuse. Rich
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
The php herd is the maintainer, so you should php-b...@gentoo.org (the herd-to-email mapping can probably be found elsewhere). Still, please get more of a feeling for how the Portage tree is put together before doing more automated bug filing, please. Cheers, Dirkjan On Wed, Oct 31, 2012 at 2:08 PM, Marco Poletti poletti.ma...@gmail.com wrote: On 31/10/2012 13:49, Dirkjan Ochtman wrote: Simply look at the metadata.xml files which can be found in each package directory. I don't find any maintainer email there... For example: ~ cat /usr/portage/dev-lang/php/metadata.xml ?xml version=1.0 encoding=UTF-8? !DOCTYPE pkgmetadata SYSTEM http://www.gentoo.org/dtd/metadata.dtd; pkgmetadata herdphp/herd use flag name='cli'Enable CLI SAPI/flag flag name='embed'Enable embed SAPI/flag flag name='enchant'Add supports Enchant spelling library./flag flag name='fileinfo'Add fileinfo extension support/flag flag name='filter'Add filter extension support/flag flag name='fpm'Enable the FastCGI Process Manager SAPI/flag flag name='gd'Adds support for gd (bundled with PHP)/flag flag name='hash'Enable the hash extension/flag flag name='json'Enable JSON support/flag flag name='ldap-sasl'Add SASL support for the PHP LDAP extension/flag flag name='mysqlnd'Use native driver for mysql, mysqli, PDO_Mysql/flag flag name='intl'Enables the intl extension for extended internalization support/flag flag name='pic'Force shared modules to build as PIC on x86 (speed tradeoff with memory usage)/flag flag name='pdo'Enable the bundled PDO extensions/flag flag name='phar'Enables the phar extension to provide phar archive support/flag flag name='suhosin'Add Suhosin support (patch and extension from http://www.suhosin.org/)/flag flag restrict=gt;=dev-lang/php-5.3.6_rc1 name='suhosin'Add the Suhosin patch from http://www.suhosin.org/)/flag flag name='sqlite2'Add sqlite2 support. Will be removed/flag flag name='xmlreader'Enable XMLReader support/flag flag name='xmlwriter'Enable XMLWriter support/flag flag name='zip'Enable ZIP file support/flag /use /pkgmetadata Marco
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
Maybe you should remember that non-devs simply /aren't allowed/ to assign bugs correctly... And if you look closer into these bugs, you might discover that jer instructed this guy to file separate bugs. (see #440178) aranea signature.asc Description: PGP signature
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On Wed, Oct 31, 2012 at 9:18 AM, li...@aixah.de wrote: Maybe you should remember that non-devs simply /aren't allowed/ to assign bugs correctly... And if you look closer into these bugs, you might discover that jer instructed this guy to file separate bugs. (see #440178) Fair points, and clearly this wasn't malicious. I think we still want to find a better solution than logging dozens of bugs for the wranglers. Perhaps where there is a need we can give elevated privs in bugzilla to those running scripts who understand what they're doing. Or non-devs could ask a dev to run the script for them after review. Rich
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 10/31/2012 08:52 AM, Tomáš Chvátal wrote: 2012/10/31 Dirkjan Ochtman d...@gentoo.org: That's rather unsurprising... If you're going to file bugs in a semi-automated manner, might as well try to assign to the correct maintainer? Yep he should've assign them, but anyway the annoying elog messages are an issue. And quite few packages suffer from it :-) Tom I disagree on most of them (and have marked the KDE-related bugs as WONTFIX appropriately). Messages that tell the user about config options, or for x functionality install y (at least until we get SDEPEND or something similar added to portage) should show up every time in my opinion. Only initial config and you just enabled (flag) really merits this. Basically, I would rather the user get too many elog messages than not enough, since I feel that a lot of people skip over them anyway and so the only display once method makes it far too easy for important messages to get lost in the shuffle. Chris Reffett -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.19 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://www.enigmail.net/ iEYEARECAAYFAlCRQQAACgkQ23laikJhg1Q+aQCeLfXsAmbtXNGOcBzM/vJaMat2 ly0AoKFzB3tPLaSO2RK0p2rWd6CoKMXm =J+3S -END PGP SIGNATURE-
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On Wed, 31 Oct 2012 13:49:37 +0100 Dirkjan Ochtman d...@gentoo.org wrote: Though if you don't know these kinds of basics, I'm not sure you should be doing *any* (semi- or not) automated bug filing. What if you don't have the privilege to assign bugs, but are willing to do the work of filing the bugs? jer
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On 10/31/2012 10:05 AM, Rich Freeman wrote: On Wed, Oct 31, 2012 at 9:18 AM, li...@aixah.de wrote: Maybe you should remember that non-devs simply /aren't allowed/ to assign bugs correctly... And if you look closer into these bugs, you might discover that jer instructed this guy to file separate bugs. (see #440178) Fair points, and clearly this wasn't malicious. I think we still want to find a better solution than logging dozens of bugs for the wranglers. Perhaps where there is a need we can give elevated privs in bugzilla to those running scripts who understand what they're doing. Or non-devs could ask a dev to run the script for them after review. This would be nice anyway. I'm generally capable of reading metadata.xml and determining whether or not to tag my bugs with e.g. STABLEREQ. One less bug to wrangle.
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On Wed, Oct 31, 2012 at 02:18:42pm +0100, li...@aixah.de wrote: maybe you should remember that non-devs simply /aren't allowed/ to assign bugs correctly... That is not correct. I am no dev and i do have edit-bugs privileges. They were given to me on request by a developer from a project i am contributing to. However, that developer *has to* review any change I do, as he/she is responsible for my mistakes. The other way is to join the bug-wranglers project if you are willing to help: http://www.gentoo.org/proj/en/qa/bug-wranglers/index.xml#doc_chap4 -- Cyprien Nicolas (fulax) Gentoo Lisp Project contrib pgpYnoZSLPKmv.pgp Description: PGP signature
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
Quoting Chris Reffett (2012-10-31 16:17:20) Yep he should've assign them, but anyway the annoying elog messages are an issue. And quite few packages suffer from it :-) It would actually be hard to find useful message which place is in elog showing up every install. Why not move such thing on Wiki for example? I disagree on most of them (and have marked the KDE-related bugs as WONTFIX appropriately). Messages that tell the user about config options, ...are *really* annoying, especially when you read it 100th time and especially the one from kdelibs: Your homedir is set to \${HOME}/.kde4. or for x functionality install y (at least until we get SDEPEND or something similar added to portage) should show up every time in my opinion. elog message for optional dependencies could be unified if must be shown every time. Only initial config and you just enabled (flag) really merits this. Basically, I would rather the user get too many elog messages than not If you assume that users time have no value and we can flood them with meaningless messages. enough, since I feel that a lot of people skip over them anyway and so the only display once method makes it far too easy for important messages to get lost in the shuffle. If user gets hundreds of such useless messages it is almost sure that he/she will miss few hidden important messages. The example could be udev long elog message which ONE time has had hidden very important message which I have unfortunately missed and ended up with unbootable system. In current form these messages have no use. I have already highlighted this problem on mailing list: Subject: [gentoo-dev] Useless messages (elog, ewarn, etc) in ebuilds Id: 20120821132457.4319.78667@localhost -- Amadeusz Żołnowski signature.asc Description: signature
Re: [gentoo-dev] [warning] the bug queue has 100 bugs
On Wed, Oct 31, 2012 at 5:54 AM, Rich Freeman ri...@gentoo.org wrote: On Wed, Oct 31, 2012 at 8:31 AM, Dirkjan Ochtman d...@gentoo.org wrote: If you're going to file bugs in a semi-automated manner, might as well try to assign to the correct maintainer? How about a policy - no automated bugs may be logged to the bug wranglers without their prior approval/review. I wouldn't think about running a script against somebody's bugzilla/repository/forum/whatever without some kind of prior communication with the administrators. This usually is considered abuse. Rich Bugzilla has a supported scripting API (via XMLRPC), and infra supports people using it, so from our POV this is legitimate. -A