Wednesday, 2009-06-10

../irclogs/#mantishelp.2009-06-10.log
--- scribe started ---00:00
mantisbotNew bug: Bug 10574 - Jimi_Joma - open - new06:58
mantisbotNew bug: Tableheader are not generated - http://www.mantisbt.org/bugs/view.php?id=1057406:58
dhx_mcan we have another look at 10330 for the 1.2 release? (would it be better if I split the renaming of constants for a later release?)08:12
giallubug 1033008:48
gialluwake up mantisbot!08:49
dhx_mhaha09:05
giallubug 1033009:05
dhx_mit was the one where I went through and removed all hardcoded uses of enum levels within the source code09:05
dhx_mand replaced them with thresholds/etc from the config file09:06
gialluah it's mantisbt.org that is dead...09:06
dhx_mit is a serious bug IMO to use hardcoded values09:06
wuehliwhat's with mantisbt.org?10:13
wuehlii cannot reach it10:13
wuehlii better use googles cache then10:14
dhx_myeah it's dead at the moment10:17
wuehlii am a total mantis newbie, just try it out at my local computer11:13
wuehlinow i wanted to change the administrator password to something else11:14
wuehliand now i cannot even log in11:14
wuehlihow can i fix this?11:14
dhx_musing administrator/root as the login?11:14
wuehlithat doesn't work sadly :(11:14
wuehliit worked two minutes ago11:14
dhx_mwhat did you do in that time?11:15
wuehlitry to change the password11:15
wuehlibut it doesn't work to log in now :(11:15
dhx_msounds like you might have mistyped the new password or something11:15
dhx_mso you might need to manually reset the password11:15
wuehliperhaps11:15
wuehlihow to do that?11:16
dhx_mare you using MySQL?11:19
dhx_mor do you have some other way to edit your database manually?11:19
wuehlidhx_m: yes, mysql11:35
AzaThtsite down?11:35
wuehlidhx_m: i am away for a bit, come back soon11:35
dhx_mAzaTht: yep11:36
AzaThtdhx_m: severe or temp?11:36
dhx_mwuehli: ok, but seeing as you just installed it, it might be easier to just reinstall over-the-top?11:36
wuehlidhx_m: okay11:36
dhx_mAzaTht: will most likely be fixed in a few hours when paul_ is about :)11:36
AzaThtheh11:36
dhx_mAzaTht: assuming it isn't a datacentre problem or something more severe11:36
AzaTht[13:37:00] -NickServ- paul_ is not registered.11:37
AzaThtpeople should remember to register alt nicks11:37
dhx_mI must admit I haven't bothered registering here yet11:39
dhx_mwas waiting for another nick to possibly become available after the FreeNode cleanup :D11:39
AzaThthehe11:40
AzaThtwhen is the purge going to take place now again11:41
dhx_mnot sure, messages said it'd be "today" or "very soon"11:41
dhx_mmight have already happened for all I know :)11:41
AzaTht# It will take place at 9am UTC on Thursday  11th June 2009.11:42
AzaThtgonna try register AzaToth then :)11:45
dhx_m./nickserv info something11:45
dhx_mwill return the status of the nick "something"11:45
AzaThtI know11:46
AzaTht /ns info AzaToth11:46
AzaTht-NickServ- Last seen  : Feb 20 15:05:25 2009 (15 weeks, 4 days, 20:39:20 ago)11:46
dhx_myep11:46
dhx_m60 days seems a very short period for deregistering nicknames11:47
dhx_mwhat if you go overseas for 3 months and don't use IRC in that time? etc11:47
AzaThtif you have an project cloak, then you are rather safe11:48
dhx_mah ok, that makes sense then11:48
AzaTht# We will try avoid expiring project cloaked user nicknames.11:48
nuclear_eclipseAzaTht: that would be nice if they ever got around to approving my project group....11:54
dhx_mhey :)11:54
dhx_mhah11:54
nuclear_eclipseI think it's been about a year now since I submitted a request for a FreeNode group, and bugged them in #support multiple times since then, and it's always "they're really buy, yours is still in the queue...."11:55
nuclear_eclipseI eventually just gave up all hope of ever getting a project registered here...11:56
dhx_mmust be a very complex and time consuming process :rolleyes: :p11:57
dhx_mdo they run a DNA check on you first?11:57
nuclear_eclipsewell, the problem is that they actually have somebody *call* you about the project, and whatever else to make sure it's legitimate, so it can't be automated in any way...11:58
dhx_mah ok11:58
dhx_mseems odd that they don't accept a statement/password provided on the official website of the project11:59
dhx_mor something to that effect11:59
nuclear_eclipseI've been tempted to just bomb their project submission form in hopes that they'll actually notice it, but I'm sure it won't help my chances, and it would take more effort =\11:59
dhx_mhah12:00
AzaThtheh12:04
nuclear_eclipsewow, how did I miss the fact that mantisbt.org was down?13:03
nuclear_eclipseI just restarted httpd there; it's back up now13:03
kenguestnuclear_eclipse: there's no nagios or other type of monitoring set up to watch it?13:05
nuclear_eclipseno, not currently13:05
nuclear_eclipseno my server :P13:06
nuclear_eclipsenot*13:06
nuclear_eclipsemight look into it though13:06
dhx_myay thanks :)13:10
nuclear_eclipsenp dhx_m ;013:10
dhx_mgot any small coding tasks that need doing? :)13:11
dhx_mI'm in the mood for doing something13:11
nuclear_eclipseyeah, fix paul_13:11
dhx_mhaha13:11
dhx_mwhat are your thoughts on ripping out the old SVN integration?13:12
dhx_mto me it sounds like it'll confuse a lot of users13:12
dhx_mand it really isn't something that I think anyone will want to maintain13:12
nuclear_eclipsedhx_m: I think we need to officially deprecate it for 1.2, to give peolpe time to get set up with the new plugin framework, and have an official plan to rip it out for 1.x13:13
dhx_mis there any sort of upgrade path needed?13:14
dhx_mI haven't used the inbuilt one before so I'm not really sure how it works13:14
dhx_mbut I imagine there is some data that needs porting across?13:14
nuclear_eclipsethe built-in itnegration is rather primitive; your VCS hook executes a CLI PHP script, passing it a text body that Mantis a) parses for related issue #s, and then b) attaches a note to each related issue containing the entire text body passed to the script13:15
dhx_mah, so there isn't really an upgrade path then13:16
dhx_mbecause data has already been lost in the old plugin13:16
dhx_ms/plugin/implementation13:16
nuclear_eclipseso afaic, there is no upgrade path other than to tell users how to configure the source-integration plugin to use the same regexes/features that the built-in integration uses13:16
dhx_mthen is there any benefit of keeping two implementations around at once?13:17
dhx_mif someone asks for help on the deprecated feature, they're not going to get assistance13:17
dhx_mand I would have thought most people would do the upgrade to the new plugin based VCS feature when going from 1.1 to 1.213:18
nuclear_eclipsethe benefit is that existing users are expecting 1.2 to continue functioning like 1.1 -- by officially deprecating the feature for 1.2, it keeps existing users happy, while letting them know that the feature is getting EOL'd, and giving them direction on where to look in the future13:18
dhx_malthough a lot of things aren't functioning like 1.1 - dates for example13:18
nuclear_eclipsedhx_m: to the users, dates function just like 1.113:18
dhx_mand I don't see anything wrong with that, as long as users are notified that upgrading to 1.2 will break X, Y and Z13:19
dhx_mah ok, so you're only talking about user-visible changes?13:19
nuclear_eclipseright13:19
dhx_mwouldn't there be an equal/greater concern for in house plugins/modifications made by people for 1.1 that no longer work with 1.2?13:19
dhx_mIMO if the changes are clearly spelled out for people looking to upgrade to 1.2, there shouldn't be an issue13:20
nuclear_eclipsetrue, but considering we've never supported anyone with a modified install anyways, I don't see how that's relevant13:20
dhx_mtrue13:20
dhx_mbut it seems a little unfair to me that we'd keep a deprecated feature and every time someone asks for support, they're told to upgrade to the new thing13:21
dhx_mit seems like someone they should of been told about when upgrading to 1.2 so they could perform the upgrade straight away13:21
nuclear_eclipsedhx_m: that's how deprecation works, and I never said we wouldn't support their use of the deprecated feature...13:22
dhx_mhmm I don't really have a problem myself (I track the master repo)13:22
dhx_mbut I thought it might make support/maintenance easier13:23
dhx_malthough I doubt anyone uses it, so the number of bugs being filed for that feature will likely be low :)13:23
nuclear_eclipsedhx_m: coming from the point of view of users that don't track Mantis development, they see that 1.2 is out, download it, and start to upgrade their site, and then find out that source integration no longer works would cause a lot of a) user frustration, and b) support requests for users thinking it's broken -- by making the deprecation part of the release announcements and so forth, it will a) let them know that it's going to be removed, and b) 13:26
dhx_mshouldn't users be reading the release notes before upgrading, so they know that as part of the upgrade, they need to switch to use the new VCS integration13:26
dhx_mI always thought the kernel API approach to deprecation was used when it was possible to use both the old and new approaches at the same time13:27
RonaldHi, is there documentation about what is possible how regarding plugins? --- Looking for chained dynamic allocation of 'custom' fields.... (Custom field 1 filled from mysql; field 2 gets allocated depending on selected value in field 1, etc)13:27
dhx_mwhereas in this case, that makes no sense... users will only use one approach (either old or new)13:27
nuclear_eclipseRonald: http://docs.mantisbt.org/master/en/developers13:27
nuclear_eclipsedhx_m: the reason it maeks sense is because source-integration is not included with mantisbt13:28
nuclear_eclipseif source-integration was a core feature, then it wouldn't be an issue -- it would just be a feature upgrade13:28
dhx_mtrue... you could always kill the project by disappearing as the lead maintainer, and Mantis would be left with nothing13:29
paul_nuclear_eclipse: what projgect group?13:29
nuclear_eclipsebut since the integration is a separate component, and because there's no immediate upgrade path, we have to be able to maintain compatibility with existing installations long enough for people to investigate and migrate13:29
dhx_mso maybe source-integration should be considered part of the official Mantis project?13:29
paul_nuclear_eclipse: I couldn't access box earlier13:30
paul_had httpd crashed?13:30
nuclear_eclipsepaul_: was trying to get a project set up for my leetcode site, where I run/work on multiple open source stuffs13:30
paul_so for 'leetcode'?13:30
Ronaldnuclear_eclipse, could it be the docs are somewhat incomplete for now ;)?13:30
nuclear_eclipsepaul_: httpd was pegging all four cores on the vps13:30
dhx_mRonald: yeah, they're a little behind13:30
paul_nuclear_eclipse: what was mysql doing?13:30
dhx_mRonald: but still contain a lot of useful/relevant info13:31
nuclear_eclipseRonald: it definitely is incomplete, but the events are the truly relevant bits, and they are completely documented13:31
nuclear_eclipsepaul_: mysql was sitting at 0% cpu, it was all apache13:31
dhx_mI wouldn't mind fixing some documentation... maybe I should try rewriting documentation for config_defaults13:31
dhx_mand syncing it with the docbook stuff13:32
paul_nuclear_eclipse: HMM13:32
nuclear_eclipsedhx_m: it's been considered making it part of core, but I'm not sure that something that full-featured/heavy-weight belongs in core...13:32
paul_nuclear_eclipse: if so, that would be 'different'13:32
dhx_mnuclear_eclipse: all I was trying to get at is having it use http://www.mantisbt.org/bugs/ and so forth would be better than relying on an external tracker13:33
nuclear_eclipsedhx_m: admin/configuration documentation is all based on 1.0.x manual, so we could certainly use some time/energy spent on reworking, updating, and such on the admin guide13:33
dhx_mnuclear_eclipse: unless of course we implement remote relationships to bugs on different trackers :)13:33
nuclear_eclipsedhx_m: you lost me now :P13:34
paul_we need to convert manual to xml docbook files13:34
nuclear_eclipsepaul_: you need to stop focusing on inconsequential details...13:34
dhx_mnuclear_eclipse: well let's say you want to create a relationship between a bug in source-integration and the plugin framework in Mantis core13:34
paul_nuclear_eclipse: well, I can't edit the existing manual ;/13:34
nuclear_eclipseit doesn't really matter whether manual is in SGML or XML...13:34
nuclear_eclipsepaul_: any text editor can edit the damn manual, it's no harder than editing HTMl....13:35
paul_the docbook editors that support wysiwyg all require .xml13:35
nuclear_eclipsepaul_: you don't need wysiwyg...13:35
* paul_ lazy13:35
Ronaldnuclear_eclipse, looks like it; bit abstract without any examples; I assume this is all for 1.2.x?13:36
dhx_mnuclear_eclipse: you could either do that normally by linking between two bug IDs on the same tracker... or do what launchpad does, and allow relationships to be created with bugs on remote trackers (maybe not even Mantis)13:36
nuclear_eclipsepaul_: the problem with XML docbook is that it requires a whole new toolchain to be investigated, set up, and tested13:36
nuclear_eclipseRonald: yes, 1.2 -- if you want examples, look at http://git.mantisforge.org :P13:37
Ronaldalright. and as for 1.2; best to go from the a3 release or something nightly?13:37
nuclear_eclipsedhx_m: yes, I would really like to have remote relationships; see http://www.mantisbt.org/bugs/view.php?id=10543 :P13:37
paul_nuclear_eclipse: I dont necessarily see that as a major hurdle13:37
nuclear_eclipseof course you wouldn't...13:37
nuclear_eclipsebut it works perfectly fine in SGML at the moment...13:38
dhx_mnuclear_eclipse: yep, that would be very nice :)13:39
nuclear_eclipsedhx_m: it's all down to the implementation now... ;)13:39
dhx_mdhx_m: to me it sounds like the relationships API needs reworking along with a new mantis_relationship_table schema13:41
dhx_mhere I am talking to myself now :p13:42
Ronaldnuclear_eclipse, did you see my last Q? ---- and finally is it likely I won't have to redo all my plugin work as 1.2 development goes along? (I have this 'fine' coworker insisting on something imho irrelevant for which i'd need all this)13:42
nuclear_eclipseRonald: yes, nightly builds are definitely "preferred" at this point; we should have been making more test releases all along....13:43
nuclear_eclipseand yes, the existing portions of the plugin system should be rather stable13:43
Ronaldexcelent :)13:43
Ronaldthanks13:43
nuclear_eclipsenp13:43
dhx_mnuclear_eclipse: example schema: relationship_id, source_bug_id, relationship_type, destination_bug_id, destination_type, destination_url, ...13:43
nuclear_eclipsedhx_m: that's why I don't think it'll be making it into 1.2 unless it's a completely standalone plugin13:44
dhx_mnuclear_eclipse: although maybe it'd be better to have a table of "remote bug trackers" so that you have a central place to change URLs, user/pass things to update the remote repo, etc13:44
dhx_mnuclear_eclipse: yep, there is no chance for it in 1.2... too much work13:45
nuclear_eclipseie, if it's fully self-contained, than any 1.2 install can use it if they want it13:45
dhx_mI'm not sure it can be fully contained, unless it completely replaces the existing relationships support in Mantis13:45
nuclear_eclipseyeah, I think that defining a set of "blessed" remote trackers/types would be the most prudent...13:45
dhx_mbut would that mean users have to first add a remote tracker before they can create a relationship with it13:46
nuclear_eclipsewhich makes sense IMO13:46
dhx_massuming of course Mantis couldn't just add it automatically on the first use13:46
dhx_myep13:46
dhx_mfrom a user standpoint, they'd just want to punch in the URL to the bug and let Mantis do the rest for them13:46
dhx_mthe admin of the local tracker can come along later and clean it up/adjust sync settings/etc later?13:47
nuclear_eclipseeg, if the plugin new about a basic set of tracker "types" (such as Trac, BugZilla, Launchpad, etc), then an admin could simply define a list of remote trackers, each of a given type with a specific base url pattern, and then the plugin can figure out the rest13:47
dhx_mwould it only be a pull system where Mantis pulls from the remote tracker? or would it also push changes?13:47
dhx_myep13:47
nuclear_eclipsedepends on how ambitious the plugin author is :P13:48
dhx_mbut what if some of those are private and need a user/pass to view it?13:48
dhx_mthat is why I think it needs to be possible to define some "sync settings" on a per-tracker basis13:48
dhx_mnot just per-tracker-type basis13:48
dhx_mthe architecture to me sounds like it should be like source-integration13:49
nuclear_eclipsewell, that's what I meant13:52
dhx_myep :)13:52
nuclear_eclipseeg, have a few base tracker types, and allow an admin to create information about a remote tracker, giving it a base type, a base url, and any other necessary info, and then users just select a remote tracker by name and tell it was the remote bug # is13:52
dhx_malthough I see victor mentioned some work he did on the SOAP feature in Mantis (on the ML)13:53
dhx_mI guess that would be something that is used where possible to talk to other bug trackers that also support SOAP13:55
dhx_mbut most trackers don't support SOAP (afaik) so it'd still be necessary to communicate via other ugly means13:55
wuehliFatal error: Cannot access private property PHPMailer::$smtp in /usr/share/mantis/www/core/email_api.php on line 80514:59
wuehliwhat can i do then?14:59
mantisbotNew bug: Bug 10575 - buknon - open - new15:02
mantisbotNew bug: autocomplete for profile with full test search in all related field - http://www.mantisbt.org/bugs/view.php?id=1057515:02
wuehlii was trying to create a new account15:02
nuclear_eclipsewuehli: what version are you using?15:34
wuehlithe version, that debian testing did provide15:36
wuehli1.1.615:37
wuehlianyway i managed to create that new account nevertheless15:38
wuehlidon't know what i did exactly :)15:38
nuclear_eclipseI think it's a problem with phpmailer that we fixed in the version we include in our releases, but I can't remember15:38
wuehliokay15:38
nuclear_eclipsemy only suggestion is to try running an official 1.1.8 release tarball from our site, and see if you still have the same problem15:39
wuehlii think as my company uses debian i stick with that debian version for now15:39
wuehlinuclear_eclipse: but thanks for answering!15:39
nuclear_eclipsedhx_m: I think you just increased the scope of #10543 by about three orders of magnitude with your comment :P15:40
nuclear_eclipsewuehli: you're welcome;  the debian maintainer sometimes idles here (aptituz), so you might try sticking around and asking him if he knows about the problem15:40
wuehlinuclear_eclipse: ah, cool15:41
nuclear_eclipseactually, aptituz_ is here right now, but not sure if he's afk15:41
wuehlii just come back to #freenode in the next days and listen to you, i might learn something :)15:44
dhx_mnuclear_eclipse: haha15:44
dhx_mnuclear_eclipse: although it is harder than it initially sounds with pushing/pulling15:45
nuclear_eclipsedhx_m: I think that push/pull balloons complexity far too large15:45
dhx_mnuclear_eclipse: I think things should only ever happen on a push basis anyway15:46
nuclear_eclipseor at least push/pull with automatic status setting, etc15:46
dhx_mnuclear_eclipse: in terms of efficiency and scaling15:46
nuclear_eclipsethe most I'd like to see is a note/notification if the remote issue changes, and otherwise leave it up to the local site/users to update the local issue as needed15:46
dhx_mnuclear_eclipse: although that is an unrealistic dream, so I guess pulling is required... even though it is an ugly approach15:46
dhx_myep I think I agree with you there15:47
dhx_motherwise it gets too complex15:47
dhx_mand you suddenly need 150 new settings to toggle for each bug tracker you link in15:47
dhx_mwhat is the closed or resolved status on the remote tracker? 2 settings15:48
nuclear_eclipseyeah, I don't like lots and lots of settings15:48
dhx_mwhat is the URL? HTTP or HTTPS? etc15:48
dhx_mthen you can get into even more fun... TLS certificates (client side, server side, etc)15:49
nuclear_eclipsesource-integration has more settings than a like, but it's the only way to do that much without assuming too much15:49
dhx_mI think it is OK myself15:49
dhx_mit's not like you have "Number of changesets to import each cycle" :D15:51
nuclear_eclipselol15:52
* paul_ pokes nuclear_eclipse 17:30
nuclear_eclipsehi paul_17:40
paul_did you see http://www.theregister.co.uk/2009/06/08/webhost_attack/ ?17:41
nuclear_eclipseouch17:42
paul_some of the flaws (http://securityreason.com/wlb_show/WLB-2009060016) e.g number 8 look quite cute ;p17:43
paul_nuclear_eclipse: did you look at the url then when your network came back last night/18:05
nuclear_eclipsepaul_: yeah, it's the same paste you gave me from the previous day?18:06
paul_probably18:06
nuclear_eclipsedisregarding function naming, I just still don't understand why you're so opposed to what I've done18:07
paul_because I think we should go through properly18:08
paul_the visible bugs are mainly fixed already18:08
paul_for example18:08
paul_atm, we we use str_pad to display a seperator under heading in changelog18:08
paul_as a question, couldn't we do that with CSS?18:08
paul_then people could format the header18:08
nuclear_eclipsewell, let's ignore roadmap/changelog; they're rubbish anyways18:09
nuclear_eclipseI just think that it's the wrong approach to use two different variants of the same function throughout the code based on the context of what's in the variable -- what's so bad about always using a utf8-capable string function, other than the very slight performance drop?  I think the added simplicity/consistency in the codebase more than makes up for that...18:11
paul_I'm more saying it's a manual job18:13
paul_for example18:13
paul_utility_api defines is_blank (is blank looks for empty string and can't use empty as empty matches (string)"0" as true18:14
paul_mantisenum.class.php calls  if ( strlen( trim( $enumString ) ) == 0 ) {18:14
nuclear_eclipsepaul_: I fixed the issue with utility api18:14
paul_the implementation of that is basicalyl what we define is_blank for18:14
paul_bah, dinner brb18:15
nuclear_eclipsepaul_: I still don't see how that case has anything to do with why my branch is "wrong" -- that's a problem with that portion of the code, and has no bearing on how we replace strlen with a utf8-capable variant...18:15
paul_right bug objects18:28
paul_and issue with utility api?18:29
paul_nuclear_eclipse: can you remind me18:32
paul_the tests stuff18:32
paul_are you able to clarify what the function should return ?18:33
nuclear_eclipseI have no clue how any of testing stuff works anymore18:34
paul_ignoring that a sec :P18:34
paul_url_sanitize('') should return index.php right?18:35
nuclear_eclipseI don't remember, and I wasn't the one that set the basic logic anyways; I only fixed some of the problems with it's parsing or whatever18:36
markw_is there any way to log what users login?18:40
markw_I see a "last login time" but is there any logging functionality built into mantisbt?18:40
nuclear_eclipsethere is debug logging for emails and filters, but no real audit logging for authentication or other things, no18:41
markw_nuclear_eclipse: yeah, trying to track down a session problem.18:42
nuclear_eclipsehowever, with the new debug logging system in place, it should be possible to extend that to the authentication modules18:42
nuclear_eclipsemarkw_: what session problem?18:42
markw_nuclear_eclipse: I think they're trying to use a shared login, but they're saying they're not.18:42
paul_expected string <abc.php?#a>18:42
paul_difference      <       xx?>18:42
paul_got string      <abc.php#a>18:42
paul_I guess that's valid18:43
nuclear_eclipseright, the ? is supelfluous/unneeded18:43
markw_nuclear_eclipse: I was looking at checking the logins and say "yeah, right here, you had 3 different systems login in the last 5 minutes"18:45
markw_nuclear_eclipse: they try to go to the "view my items" page and it sounds like it kicks them back to the login.18:46
nuclear_eclipsemarkw_: with mantis, that should actually work just fine, assuming that they're using the correct username/password18:46
markw_using the same username/password?18:46
nuclear_eclipseare they getting an error about an "invalid session"?18:46
paul_nuclear_eclipse: is the expectation we encode array( 'plugin.php?page=Source/list&id=1 to %2F ?18:47
nuclear_eclipsepaul_: I don't know18:47
paul_well, you modified it ;/18:47
nuclear_eclipselies18:47
markw_nuclear_eclipse: problem is that it's developers in japan trying to read what the fieldtesters in the US are submitting.18:48
markw_Here's the description of the "problem"18:48
markw_This message is displayed as soon as changing to other ID.18:49
markw_However, it is possible to change to a detailed page on each page once every 3018:49
markw_times.18:49
markw_And other members except me(they have already registered) cannot access the page18:49
markw_of My view page.18:49
paul_& in url's should be encoded &amp; ?18:50
markw_so there's an japanese<>english translation problem going on here.18:50
nuclear_eclipsemarkw_: I'm not sure what to say then =\18:51
markw_nuclear_eclipse: neither am I, since it works fine for the 50+ other users in the system.18:51
markw_except they're not behind a corporate firewall in japan.18:51
nuclear_eclipsemarkw_: what version are you using?18:54
markw_1.218:54
nuclear_eclipsehow new?18:54
markw_hmm.18:54
markw_a3?18:54
nuclear_eclipseeg, official 1.2a3 release, or something newer from git/snapshot?18:54
markw_official18:54
markw_we needed some of the fancy exporting stuff.18:55
markw_meeting...18:55
markw_bbl18:55
nuclear_eclipseyou could try setting $g_session_validation = OFF in your config_inc18:55
markw_ok, will try.18:55
markw_but that could introduce security risks right?18:55
markw_japan doesn't come back online until 4-5pm18:56
nuclear_eclipseif you were running latest snapshot, it actually allows individual users to disable validation for their own session, and would allow most users to retain the security18:56
markw_anyway, have a noon meeting.18:56
nuclear_eclipsecheers18:56
markw_hmm.18:56
markw_I'll have to look at that.18:56
markw_bbl18:56
paul_nuclear_eclipse: what do you think the purpose of string_sanitise_url is?18:57
nuclear_eclipseno freakin clue18:57
paul_but you fixed it last!18:57
paul_nuclear_eclipse: are you able to fix the function for me?19:05
nuclear_eclipsethat doesn't mean I understand *why* we use the function, I just understand what it's doing and how to make it behave...19:05
paul_expected string <abc.php#a>19:05
paul_difference      <       xx?>19:05
paul_got string      <abc.php?#a>19:05
paul_could you fix that then? :)19:05
paul_and you strip <script> from urls?19:08
paul_and do we escape " in query strings to \" then urlencode that?19:12
nuclear_eclipseI don't know anymore; it's been a long time since I touched that...19:12
nuclear_eclipseI think right now is hardly the time to bring up random stuffs; you still need to fix/commit bug objects and due dates....19:13
paul_it's not really that random19:13
paul_it means my automated cc stuff fails every build19:14
paul_which means the scripts i've got to pick up formatting errors dont run19:14
nuclear_eclipseso stop cc from checking the url tests for the time being...19:15
nuclear_eclipseit's really not that hard of a concept ;)19:15
paul_if ( !empty( $t_clean_pairs ) ) {19:19
paul_$t_query = '?' . join( '&amp;', $t_clean_pairs );19:19
paul_}19:19
paul_shouldn't we check it's not empty before adding ? ?19:19
RoellI have a question19:36
Roellis Git supported with Mantis?19:36
Roelllike Svn is19:37
nuclear_eclipseRoell: depends on how you mean19:38
nuclear_eclipsefor 1.1.7, any VCS is supported that can use a commit hook to execute the checkin script19:38
nuclear_eclipse1.1.x*19:38
nuclear_eclipsefor 1.2.x, there is a plugin available called Source Integration (used on mantisbt.org) that has a lot more functionality, and is currently compatible with both SVN and Git19:39
nuclear_eclipsefor the latter option, see http://leetcode.net/blog/2009/01/integrating-git-svn-with-mantisbt/19:40
nuclear_eclipsefor the former, see the mantisbt manual at mantisbt.org19:40
Roellgee thanks!19:47
nuclear_eclipsehmm, paul_ have you gotten any git mails?19:51
nuclear_eclipsejust realized that the last two commits haven't triggered mails or CIA-6119:52
nuclear_eclipsepaul_: httpd is pegging cpu again...19:52
nuclear_eclipseserver's still responding though atm, so can you investigate?19:53
nuclear_eclipse~10 httpd processes are together pulling 100% on all four cores19:54
nuclear_eclipsepaul_: getting a bunch of these in the log:19:58
nuclear_eclipse[Wed Jun 10 15:57:53 2009] [error] [client 65.55.106.242] PHP Fatal error:  Maximum execution time of 30 seconds exceeded in /var/www/html/bugs/core/lang_api.php on line 10419:58
markw_nuclear_eclipse: any db changes between the latest and 1.2.0a3?20:10
nuclear_eclipsemarkw_: lots and lots; we rearranged the way dates worked -- if you're actually using the due date field, latest git is currently broke on that, but otherwise it should be fine after you do a schema upgrade -- *highly reccommend* that you back up your database and installation before upgrading, just in case20:13
markw_yeah, I snapshot everything nightly.   I'd do a dump beforehand.  I'll let the field test manager know that it may solve our problem.20:14
nuclear_eclipseok20:14
paul_nuclear_eclipse: sure20:21
paul_nuclear_eclipse: have you updated mantis on live site?20:22
nuclear_eclipsepaul_: it's running the 'live' branch from git, which is master from mid-may (before your date changes) with a couple small changes cherry-picked20:24
nuclear_eclipsedo `git log live` on your local to see20:24
paul_msn search ;/20:26
paul_mid may?20:27
nuclear_eclipsesorry, mid-april20:28
paul_before or after 13th?20:29
nuclear_eclipseaccording to `git log live`, last commit from master before cherry-picking started was:20:30
nuclear_eclipsecommit 085f672d06d774d1e2d4bffe093bf7e2579fb2fb20:30
nuclear_eclipseAuthor: Siebrand Mazeland <s.mazeland@xs4all.nl>20:30
nuclear_eclipseDate:   Wed Apr 15 09:36:57 2009 +020020:30
nuclear_eclipse    Localisation updates from http://translatewiki.net (2009-04-15 07:04 UTC)20:30
paul_anyway, it's msn trying to word export everything20:31
nuclear_eclipseah...20:31
paul_probably cause coreformatting plugin could do with being optimised20:31
nuclear_eclipseor we could add rel="nofollow" to links like that...20:31
markw_nuclear_eclipse: just curious, if they were sharing a login on the other end and they were using the same proxy to access mantis, would that cause session issues?  Session data looks to be tracked/stored by ip in php's session dir.20:32
nuclear_eclipseno, mantis should handle simultaneous login on the same account just fine20:33
nuclear_eclipsePHP tracks session by cookie ID20:33
markw_I know it did funny things with two different instances of mantis, but that was because of a session name variable or something.20:33
nuclear_eclipseyeah, between  two installs of mantis on the same domain, you need to set them to have different cookie prefixes20:34
markw_yeah, found that out.  but this is the same instance.20:34
nuclear_eclipseyeah, in that case, the problem with proxies is when the proxy has multiple possible exit nodes with different IP addresses, in which case the session validation in mantis suspects a session hijack -- setting $g_session_validation = OFF or (in latest git) having the user uncheck "Secure Session" at login will disable that validation20:36
markw_I wonder if something is being filtered at the proxy, I could see corp doing some sort of filtering.20:37
nuclear_eclipsepaul_: http://git.mantisforge.org/w/mantisbt/jreese.git?a=shortlog;h=refs/heads/utf820:40
nuclear_eclipsepaul_: http://git.mantisforge.org/w/mantisbt/jreese.git?a=commitdiff;h=8a523c480ffa7f2b97537d8695d0bac3116e3ea4;hp=265e689944698b99d412c5b14b0b5d2b366e17e420:41
paul_nuclear_eclipse: I still think we should do it gradually :P20:53
paul_albeit, plugins etc should use the public functions20:53
nuclear_eclipsepaul_: we don't have time for gradual, and I still don't think we should have half the core using a different variant from the other...20:53
nuclear_eclipseI don't even understand what the point of that would be, other than  performance at the expense of consistency...20:54
paul_for example20:55
paul_http://git.mantisforge.org/w/mantisbt/jreese.git?a=commitdiff;h=8a523c480ffa7f2b97537d8695d0bac3116e3ea4;hp=265e689944698b99d412c5b14b0b5d2b366e17e4#patch1820:55
paul_-                                       header( 'Content-Length: ' . strlen( $t_dot_output ) );20:55
paul_+                                       header( 'Content-Length: ' . utf8_strlen( $t_dot_output ) );20:55
paul_if that content-length is binary object, do we know utf8_strlen is safe to call on it20:55
nuclear_eclipseI don't see how that would mean anything...20:56
nuclear_eclipseif it's binary data, then there won't be any utf8 encoding, so it'll work just like strlen....20:57
paul_I still think your partly missing the point ;/21:00
paul_changing strlen->utf8_strlen on it's own is unlikely to fix utf8 bugs21:00
nuclear_eclipseI think you're focusing too much on picky details ;)21:00
nuclear_eclipsefacepalm21:00
paul_well, i've grepped for strlen here21:00
paul_before21:01
nuclear_eclipselike I've told you before paul_, this is only the first step....21:01
paul_yep21:01
paul_my point is i've grepped for all the non-utf8 php functions we have here21:01
paul_printed them out21:01
paul_and started to highlight which ones require utf8 and which ones dont21:01
paul_and what ones need fixing to support utf821:01
nuclear_eclipsepaul_: my point is that devs shouldn't have to look at each use and decide if it needs utf8 support or not!21:02
paul_where is_blank gets called 7000 times on some pages21:02
paul_i'd rather the faster version personally21:02
nuclear_eclipseif we just use a utf8-capable function across the board, it makes sure we never have a problem...21:02
nuclear_eclipsepaul_: if the string is not utf8-encode, mb_strlen is just as fast as strlen....21:03
paul_the compat version isn't21:03
nuclear_eclipsearg21:03
paul_i've spent quite a chunk of time looking at utf8 stuff over last few months21:03
nuclear_eclipseperformance is not the only thing to worry about anyways21:04
paul_for example21:04
nuclear_eclipsepaul_: you've spent quite a chunk of time and got nothing to show but some highlighted pieces of paper; I spent an afternoon and have a patch that fixes three functions across the board, with a path for continuing the work for any other string function21:04
paul_http://git.mantisforge.org/w/mantisbt.git?a=commit;h=49ca189e400a20fb4dc9ea64c61525a2cf3c5e9221:05
nuclear_eclipseok, what does that have to do with anything?21:05
paul_those sort of commits are actually fixing issues21:05
paul_it takes about 10 seconds to string replace strlen -> utf8_strlen21:06
nuclear_eclipseyes, but that has nothing to do with utf821:06
paul_actually it does21:06
paul_as wordwrap doesn't support utf821:06
paul_but the phpmailer code to wrap text appears to21:06
nuclear_eclipseso that's just the next step for utf8_api....21:06
paul_no, we dont use the wordwrap function anywhere in mantis anymore21:07
nuclear_eclipsethen who cares?  you're putting up these "examples" that don't actually make a point as to why my approach is incorrect...21:07
paul_i've already said i'd rather the performance increase of using strlen for core api functions where we dont require utf8 support21:08
nuclear_eclipsethat just makes the core apis harder to understand and work with though, and I'm not sure why we should lose that for some very slight performance gains....21:09
nuclear_eclipseand btw, is_blank shouldn't be using strlen anyways -- it only needs to know if the string contains any characters at all...21:10
paul_!empty isn't any faster21:10
paul_(as if str = "0"; then empty(str) == true, and if you add the || str == "0" check, the speend gain of empty balances out)21:11
paul_unless I'm missing something21:11
nuclear_eclipseis_string() && isset( $s[0] )` gets around the empty('0') issue, and doesn't require counting....21:11
nuclear_eclipsestrlen() is linear time based on the length of the string, versus isset( $s[0] ) is constant time...21:12
nuclear_eclipseeg, if you have a string (like description field) that has a really long string in it, strlen() will run considerably slower than isset( s[0] )21:13
nuclear_eclipseanywho, time to head home again...21:14

Generated by irclog2html.py