Monday, 2009-06-01

../irclogs/#mantishelp.2009-06-01.log
--- scribe started ---00:00
GuillermoRamirezhello06:15
MaximBhi people06:15
MaximBI REALLY need your help please306:16
GuillermoRamirezI am a new user06:16
MaximBI've installed latest stable mantis , when I go to the /mantis/admin/install.php page it goes fast , but when I go to the login page it's really slow and unworkable , I even disabled SSL but it's still slow - where can be the problem ?06:17
GuillermoRamirezI need help06:20
GuillermoRamirez how can I create a roadmap06:20
MaximBdo you ever help in this channel ?07:07
GuillermoRamirezhello07:44
nuclear_eclipsemorning all12:30
[KK]Kirillnuclear_eclipse: lo12:35
portej05hi12:54
[KK]Kirillportej05: lo12:57
portej05hi, I've got a question about mantis, but I'm not an administrator nor do I have access to the configuration of the current setup12:59
portej05We're running mantis 1.0.5 (fairly old)12:59
portej05Thing is, it's sending out emails for everything, except account activation12:59
[KK]Kirillsorry, I forgot this version12:59
nuclear_eclipseportej05: my only recommendation is to upgrade to 1.1.7 (latest stable), which should not only fix that, but also fixes many security issues that were present in 1.0 and early 1.113:04
portej05Cheers nuclear, I'll pass that on13:05
portej05Is the upgrade process relatively painless?13:06
nuclear_eclipseassuming you don't have any modifications to the code (only changes to custom_* and config_inc), then it should be as simple as extracting the new code over top the old, and visiting /admin/install.php to run the upgrade procedure13:07
portej05Thanks!13:09
* daryn_away is away: Gone away for now14:39
* daryn_away is away: Gone away for now14:39
* daryn_away is away: Gone away for now14:39
* daryn_away is away: Gone away for now14:48
* daryn_away is away: Gone away for now14:57
nuclear_eclipsepaul_: you about?15:29
paul_nuclear_eclipse: lo17:48
paul_nuclear_eclipse: what have i broken? ;/17:48
nuclear_eclipsenothing, wanted you to review some stuff for 1.2 so we can get a release out by the weekend17:53
mantisbotNew bug: Bug 10548 - irwansun - open - new17:54
mantisbotNew bug: reporter only able to choose 2 severity and 1 reproducibility when reporting an issue - http://www.mantisbt.org/bugs/view.php?id=1054817:54
paul_nuclear_eclipse: sure17:54
paul_you dont want to know what i'm working on17:54
nuclear_eclipseas usual17:54
paul_anyway what you wanted me to look at?17:55
nuclear_eclipsejust a sec17:56
paul_also seepm18:04
* daryn_away is away: Gone away for now18:08
nuclear_eclipsepaul_: http://git.mantisforge.org/w/mantisbt/jreese.git?a=shortlog;h=refs/heads/http abstracts http caching headers into a separate API for use by plugins -- just verify that caching still works correctly on your end after the change18:15
nuclear_eclipsehttp://git.mantisforge.org/w/mantisbt/jreese.git?a=shortlog;h=refs/heads/project-names changes the way project/category names are displayed in a bug and on the my view page -- hopefully to fix the complaints by users and my employer that you can't tell what project a bug is from anymore, now that categories don't always share the same project as they bug they're attached to18:16
nuclear_eclipsehttp://git.mantisforge.org/w/mantisbt/jreese.git?a=shortlog;h=refs/heads/plugin-uses may not be testable by you, but it adds the ability to specify "soft" dependencies between plugins, which won't prevent install/usage if the soft dependency is missing, but handles correct init order if it is there18:18
paul_how would plugins use it?18:28
nuclear_eclipseby setting a ->uses property in the same method as it for ->requires18:28
paul_but your calling http_all_headers from core.php?18:29
nuclear_eclipseoh, soryr18:29
nuclear_eclipsewas thinking of the last branch18:29
nuclear_eclipsefor headers18:30
paul_I half dislike the concept of adding an api called http for sending headers though ;/18:30
nuclear_eclipsecore.php calls http_all_headers by default, and then a plugin page can optionally call http_caching_headers() to replace/override the default header data18:30
nuclear_eclipseand I used http_api because headers are part of the HTTP spec, and didn't really seem to fit elsewhere, and I felt that a header_api was too specific/limiting18:31
paul_i'm just thinking what we'd call a http client api18:31
paul_I think i probably agree to changing custom_headers to _global18:34
paul_although, we should update config api to reflect that change18:34
nuclear_eclipsehmm?18:34
paul_well, if we can only set it globally, we should flag you can't set it in db18:35
nuclear_eclipsewe already do..18:35
paul_oh ;/18:35
nuclear_eclipse:P18:35
paul_but this should be a null change18:37
paul_my only thought would be18:37
paul_do we need 4 functions for this18:37
nuclear_eclipseI think so18:37
nuclear_eclipseeg, if I just want my plugin page to override the caching headers, it shouldn't need to call something that will also reload and reset content-type and other custom headers....18:38
paul_ok the caching one fine18:39
paul_i'm more thinking about:18:39
paul_+function http_content_headers() {18:39
paul_+       if ( !headers_sent() ) {18:39
paul_+               header( 'Content-type: text/html;charset=utf-8' );18:39
paul_+       }18:39
paul_where that's *always* going to get called (by core.php)18:39
paul_i.e. users/plugins/whatever can't call http_content_headers or it's pointless to call it as it's already called by core.php18:40
nuclear_eclipsewell, yes, it seems superfluous atm, but I did that with the thinking that we could eventually use them as a unified way to manually set those types of headers, eg, like for file_download.php or plugin_file.php....18:40
paul_personally, I find header() a lot easier to understand (I dont need to grep the source to work out what that does)18:41
paul_that would be my only thought -> I'd like to get to the point where we've got a documented interface on functions for plugin developers to use18:41
nuclear_eclipsewell, I would generally agree; my point was simply that an API for certain things could be useful, eg, always append 'charset=utf-8' to any text/* content-type18:42
paul_adding effectively null/private functions that just call header() seems a bit silly18:42
nuclear_eclipsepaul_: the documented interface is the entire API :P18:42
paul_other then the fact i'd probably be inclined to set a @private tag against 3 of those functions ( aka custom/content/all) and just leave caching as @public18:43
nuclear_eclipseand I had the idea that http_content_headers() could eventually take a parameter for the type, and do some other processing, like the charset= type of thing18:43
paul_anyway, it looks fine18:43
nuclear_eclipsebut it's just an idea; I'm fine rolling it into the other functions if you prefer18:43
paul_but personally, i'd feel http_content_headers( 'Content-type: text/html;charset=utf-8');  as more confusing18:43
nuclear_eclipsewell, I wasn't meaning *that*....18:44
paul_as i'd need to waste 5 minutes working out what http_content_headers does to find it just calls header(); :)18:44
nuclear_eclipseI was meaning http_content_headers( 'text/plain' )18:44
paul_anyway, commit it / do whatever as it's a null change18:44
paul_that's effectively the same :P18:44
nuclear_eclipseand then the API would handle anything appropriate based on the actual content-type, such as adding other appropriate headers, etc18:45
paul_you'd need to take in content-type as one param, charset as a 2nd param18:45
paul_as for things like file_upload, its not necessarily going to be utf818:45
nuclear_eclipsewell, assuming that any text/* type is natuarally going to be utf8, and anything binary is not,....18:45
paul_and the point is one's still effectively just calling headers18:45
paul_the one thnig that patch does18:46
paul_anyway, it's a null change from what I can tell18:47
nuclear_eclipseok18:47
paul_and we've spent way too long discussing it :P18:48
nuclear_eclipselol18:48
paul_incidently feel free to add @return null to each function :P18:49
* nuclear_eclipse stabs paulr in the eye18:49
paul_so we're putting projectname in [ ] before bug id?18:51
nuclear_eclipseI couldn't think of a better place to put it that was obvious when glancing at the bug.... =\18:51
paul_but your not normally going to ever see it right?18:52
nuclear_eclipseputting it front of the category name is now ambiguous18:52
nuclear_eclipseright18:52
paul_as normally current_project will be project bug is in18:52
paul_(in fact, is that always true?)18:52
nuclear_eclipseyou'll only see it if the bug is from a project other than the one you have selected18:52
paul_the project names thing I think looks ifne too18:56
nuclear_eclipseok18:56
paul_be interesting with a long project name + summary name18:56
* paul_ is just reading diff's18:56
paul_ahh yes18:57
paul_autoloads18:57
paul_it's not possible to do late loading of plugin's is it?18:58
nuclear_eclipseno, not if we have eventual plans for pluggable authentication18:58
paul_well, it would obviously have to be clever :P18:58
nuclear_eclipsewhat are you thinking about that you would need late loading of plugins?18:59
paul_if you had 20 plugins regesitered for example18:59
paul_and a plugin that say... sent an msn message when a bug was added18:59
paul_you'd only need to load that plugin when trgierring the event18:59
paul_(I guess i'm thinking alsmost bootstrapping ?19:00
nuclear_eclipsewell, the problem is how do you know that a plugin only needs that one event if you don't load it? :P19:00
nuclear_eclipsehonestly, I think the biggest problem with loading plugins (the loading part) is completely negated by using a php opcode caching system, like what we use on mantisbt.org19:01
paul_nod19:01
nuclear_eclipsebut seriously, it's not possible to do "late loading" without somehow loading a plugin to figure out what events it hooks into; at which point it's much simpler to just load them all at once19:02
nuclear_eclipsereally, the way I've implemented them is probably as close as possible to that optimal solution, in that once a plugin is loaded, it's only ever touched if a hooked event is triggered, or if a page/file for that plugin is requested19:03
nuclear_eclipsethe overhead for an individual plugin is sizeof(PluginClass) + ~5-10 string/int pointers in the init system + 2/3 plugin/function names for each event hooked19:04
* daryn_ is back.19:05
nuclear_eclipseI can't remember the specifics on overhead atm, but it's generally pretty slim19:05
* nuclear_eclipse asks daryn to set his IRC client to *not* announce his here/away status....19:05
nuclear_eclipsethat's what /away and /back are for....19:06
nuclear_eclipsedaryn: in some channels, that would get you kicked :P19:07
nuclear_eclipsepaul_: so am I go on those three branches?19:09
mantisbotNew bug: Bug 10549 - shobana - open - new19:09
mantisbotNew bug: not able to set the client Validation. - http://www.mantisbt.org/bugs/view.php?id=1054919:09
paul_nuclear_eclipse: can you look at my graph changes later in week?19:09
paul_review19:10
nuclear_eclipsearg, I really wish people would stop reporting bugs against old versions, esp when it's bugs that have already been reported and fixed....19:10
nuclear_eclipseand yes, I'll review if possible; I've never used the graphs though19:10
paul_and i'm fine with the plugin change - i'm sure it makes sense19:11
paul_*HOWEVER*19:11
paul_i've got some things i'd like to think about with plugins19:11
paul_1) do we let you specify order for plugins to load in atm?19:11
paul_(i assume we do?)19:11
nuclear_eclipse1) yes, priority setting on the plugin management screen19:12
paul_secondly19:12
nuclear_eclipse1a) yes, combination of requires and uses flags19:12
paul_i'd like to add some validation rules to plugins or types or something19:12
paul_such that you *must* have at least 1 formatting plugin registered19:12
paul_i.e. you can't remove mantiscoreformatting without replacing it with something else19:13
nuclear_eclipseI think that starts to put too much proprietary validation into the plugin system -- not to mention that just because MantisCoreFormatting is enabled doesn't mean it has to actually be working; case in point being that you can optionally disable any or all of the coreformatting components, specifically to be compatible with other formatting plugins, but can also be done manually as well....19:15
nuclear_eclipsepaul_: so am I go on those three branches?19:16
paul_My logic is if you turn it off/disable something, you've got a security risk unless you replace with something else19:16
paul_I guess19:16
paul_i'm trying to decide whether to make graphs a plugin19:16
darynnuclear_eclipse: :P19:17
daryndone i think19:17
nuclear_eclipsethank you :)19:17
paul_daryn's back :P19:17
nuclear_eclipseno offense, but I'd rather not see channel chatter every time you go away or come back; none of us are really that important to know exactly where they are :P19:18
darynnone taken. not sure when/how that got set. don't really remember doing so.19:18
nuclear_eclipselol19:18
paul_either of you use spl_autoload stuff in php?19:18
nuclear_eclipsenope19:19
darynnope19:19
nuclear_eclipseI've never really used any of the SPL19:19
CIA-61Mantisbt: jreese * r2206801e200a / (3 files in 2 dirs): Improve performance by using string_display_line.19:19
CIA-61Mantisbt: jreese * rfcccd8275930 /core/ (classes/MantisPlugin.class.php plugin_api.php): Implemented initial support for soft dependencies with the new ->uses property.19:19
CIA-61Mantisbt: jreese * rf9396639fb11 / (4 files in 2 dirs): Reimplemented the ability to view an issue's project name alongside the category's project name.19:19
CIA-61Mantisbt: jreese * r6ab1b6defcbb / (core.php core/http_api.php): Extracted HTTP header routines into new API, allowing plugins to set caching headers for their own pages and content.19:19
CIA-61Mantisbt: jreese * rb128f30f666e /core/category_api.php: Only display category's project name if different from current.19:19
CIA-61Mantisbt: jreese * r61096547d19a / (7 files in 3 dirs): Merge branches 'http', 'plugin-uses' and 'project-names'19:19
paul_anyway back to graphs19:20
paul_grr19:20
paul_some git has changed trunk19:20
nuclear_eclipse:P19:20
darynrebase19:20
nuclear_eclipse`git pull --rebase`19:21
paul_fetch does pull+rebase right?19:21
nuclear_eclipseI just used my ~/.gitconfig to map `git pullr` to `git pull --rebase19:21
nuclear_eclipseno, fetch just does fetch19:21
nuclear_eclipsepull = fetch + merge19:21
nuclear_eclipsepull --rebase = fetch + rebase19:21
paul_fetch + rebase = pull?19:21
paul_from wht i can tell19:22
paul_the /graphs directory is dead19:22
paul_think i might make some db schema updates too19:23
paul_erm wtf19:25
paul_SELECT c.name AS name, COUNT(name) as count FROM mantis_bug_table JOIN mantis_category_table AS c WHERE c.project_id=? GROUP BY name ORDER BY name19:29
paul_what should that query be with new category api ?19:29
nuclear_eclipsewhat did that even do under the old category api..../19:36
* paul_ shrugs19:38
paul_count of categorys?19:38
paul_oh I just remembered19:49
paul_we need to fix soap api before release19:49
paul_also can you make the url tests pass :P19:49
nuclear_eclipsewhat url tests?19:51
* paul_ STARES19:51
paul_the ones i've been asking you to fix for 2 weeks!19:51
paul_in tests dir19:52
nuclear_eclipsethe ones I've been asking you for two weeks to clarify.....19:52
* nuclear_eclipse is not sure what has/hasn't changed recently with all the stuff you/giallu did to tests...19:52
nuclear_eclipsenot to mention that *all* the test pages end up outputting a blank page for me....19:53
paul_http://www.mantisforge.org:8080/cruisecontrol/buildresults/mantisbt?tab=testResults19:54
paul_just fix the failures :P19:54
nuclear_eclipseie, could use some documentation on how all the tests work, so I can actually run the tests...19:54
nuclear_eclipsepaul_: that page doesn't even tell me what is wrong, other than a failure is found...19:55
paul_Failure:19:56
paul_testStringSanitize(Mantis_StringTest) with data set #0 ('')19:56
paul_Failed asserting that two strings are equal.19:56
paul_expected string <>19:56
paul_difference      <?????????>19:56
paul_got string      <index.php>19:56
paul_/home/cc/cruisecontrol-bin-2.8.2/projects/mantisbt/source/tests/Mantis/StringTest.php:3919:56
paul_sure it does19:56
nuclear_eclipseright, that doesn't really help me at all....19:56
paul_apparently test expect '' and got 'index.php'19:56
nuclear_eclipseand I can't fix it if I can't run the tests on my own system...19:57
nuclear_eclipseand how do I even know that the tests are correct?19:57
paul_you use phpunit right?19:57
paul_or can do19:57
paul_get phpunit from pear19:57
nuclear_eclipsearg19:57
paul_pear channel-discover pear.phpunit.de19:57
paul_pear install phpunit/PHPUnit19:58
* nuclear_eclipse can't use pear on his locked-down workstation....19:58
paul_http://www.phpunit.de/manual/current/en/installation.html19:58
paul_do it manually then :P19:58
nuclear_eclipseand why am I the one tasked with fixing those failures?19:59
paul_:)19:59
paul_just fix em20:00
paul_:P20:00
nuclear_eclipseright....20:00
paul_giallu couldn't work out what url tests are supposed to do20:00
nuclear_eclipsewhere the hell do we even define what's expected?20:01
paul_but yes20:01
paul_that should be a array20:02
paul_in->out i'd say20:02
nuclear_eclipsepaul_: thoughts on issue #10217?20:12
nuclear_eclipsebug 1021720:12
mantisbotBug 10217 - kec161 - open - new20:12
mantisbotAdded ability to have the html buttons bar at the top, bottom or both of the bug view page. - http://www.mantisbt.org/bugs/view.php?id=1021720:12
nuclear_eclipsemy employer has requested this as well, and I think it would be quite useful20:12
* paul_ shrugs21:13
paul_i'm not sure you'd want both21:13
paul_how do i move files in git21:13
paul_in fact, i know this21:14

Generated by irclog2html.py