Friday, 2009-06-05

../irclogs/#mantishelp.2009-06-05.log
--- scribe started ---00:00
mantisbotNew bug: Bug 10563 - informatyk - open - new10:30
mantisbotNew bug: After customize view_issues_page_columns, the priority column crashes - http://www.mantisbt.org/bugs/view.php?id=1056310:30
mantisbotNew bug: Bug 10564 - Jimi_Joma - open - new10:40
mantisbotNew bug: Mantis is generating empty spaces in field description, summary (all textareafields) - http://www.mantisbt.org/bugs/view.php?id=1056410:40
[KK]Kirillpaul_: lo11:47
nuclear_eclipsepeople who use word documents to post screenshots should be shot on sight...13:37
* nuclear_eclipse starts upgrading his server to Jaunty15:23
cudellHI all. Got a question about a bug I think we've found in the latest nightly build.16:20
cudellCreating a user seems to fail with 500 Internal Server Error16:22
cudellthe post request seems well formed... permissions on the script seem to be in order. the response to the post is empty.16:27
markw_hmm.16:28
markw_is there any "bulk" user adding tool?16:29
cudellresending the same request again via a refresh results in a Mantis Application Error 280016:29
cudellwe are not using a bulk user adding script16:29
cudellsimply: manage_user_create.php16:30
markw_no16:30
markw_I need to bulk add some users.16:31
markw_I thought there was a "bulk" add script but it doesn't look like it.16:31
cudellit looks like the email validation in that post is broken. email_ensure_valid - line 433 of user_api.php in Core16:40
nuclear_eclipsewhat version of the nightly have you been testing?17:01
jlebHi eerybody.  I'm trying to get BT1.2 integrated with Subversion using the Source Control Integration plugin.  Not having much luck, any one here who might be able to help?19:02
nuclear_eclipsewhat's the problem?19:03
cudellhi jleb19:11
cudellthis is the nightly build version we've got: 1.2.0a3 2009-06-04-master-2b2ec0119:12
cudellSchema Version 17119:12
paul_nuclear_eclipse: .19:36
nuclear_eclipsehi paul_19:36
paul_lo19:36
cudellcan you guys verify that the email validation in that post is broken. email_ensure_valid - line 433 of user_api.php in Core19:52
paul_lo daryn19:55
paul_nuclear_eclipse: need you to review my stuff19:56
paul_tomorrow19:56
nuclear_eclipsecudell: I think it might be19:57
nuclear_eclipsepaul_: have you fixed dates yet?19:57
paul_i can't have broken custom field dates19:59
nuclear_eclipseok19:59
nuclear_eclipsebut you *did* break the ability for users to input dates for due_date, version release dates, etc19:59
paul_yep20:00
paul_shouldn't versoin release dates show calendar?20:00
nuclear_eclipseI think that's rather beside the poitn20:05
* paul_ wonders how one can apply admin checks to plugins20:07
nuclear_eclipsefacepalm20:07
nuclear_eclipsedates!20:07
nuclear_eclipsefix dates!20:07
nuclear_eclipsedon't make me revert every change to dates in order to release 1.2 RC20:08
vb1234nuclear_eclipse: hi20:08
nuclear_eclipseti Victor20:08
nuclear_eclipsehi*20:08
vb1234are you talking to paul_ ?20:08
nuclear_eclipseyep20:08
vb1234so what's up with the rc.20:09
nuclear_eclipseassuming that Paul actually gets around to fixing the new date handling before tomorrow evening, I'm going to try to roll the release tomorrow evening and/or sunday20:09
paul_+spl_autoload_register( '__autoload' );20:10
paul_that should be safe right?20:10
nuclear_eclipseno clue20:12
nuclear_eclipsewhy are you screwing around with SPL?20:12
paul_to allow plugins to call autoload functions :P20:13
CIA-61Mantisbt: jreese * rc6c18c57e4a6 /core/email_api.php: Partially revert 209070a7: "Remove old email_api functions...".20:13
nuclear_eclipsewhy would plugins need to call autoload?20:13
paul_whats up with changing the mail regex's20:16
nuclear_eclipsethe old version was slow/clunky/broken in some cases, and a simpler regex still covers 99.99% of all users20:16
* paul_ has a headache20:17
paul_someone needs to fix tests20:17
nuclear_eclipseeg, rfc comments and domain literals are worthless and/or never used20:17
paul_btw20:17
nuclear_eclipsepaul_: you need to fix dates before you start finding other things to fix....20:17
paul_as it's stopping me from going through my automated test results20:17
nuclear_eclipseseriously, dates is higher priority atm20:18
vb1234We should not have trunk at a broken state at any time.20:18
vb1234that's why we moved to git :)20:19
nuclear_eclipselies20:19
vb1234?20:19
nuclear_eclipsesarcasm20:19
vb1234:)20:20
nuclear_eclipsethe problem in this case is that paul_ developed in his public repo/branch, and everything looked fine, and then I found the breaks after the fact when testing something else20:20
paul_the problem is more20:20
vb1234then I agree with the focus on this, specially that this is a feature that affects the db schema.20:20
nuclear_eclipseeg, we were using git properly -- we're just not testing enough before merging features to master20:20
paul_it gets confusing if you have lots of patches affecting different things20:21
nuclear_eclipseand tbh, I think our biggest problem is lack of a) developer time, and b) dedicated testers for looking at feature branches20:21
vb1234When was this checked in?  From memory, it was sometime ago.20:22
vb1234correct?20:22
nuclear_eclipselack of dev time results in just pushing to master because there's never enough time to thoroughly review/test20:22
paul_nuclear_eclipse: actually one of the reasons I wanted to get dates out the way20:22
nuclear_eclipseabout middle May was the merge20:23
paul_is I was trying to test changes to bugdata + dates together20:23
paul_and having two seperate branches was hindering that20:23
vb1234can't you merge to a third branch?20:23
vb1234or some local branch?20:23
vb1234I feel we are stuck now that it is even not easy to revert.20:24
vb1234given the schema change.20:24
nuclear_eclipsevb1234: I suspect paul_ is averse to learning how his tool works20:24
vb1234and us / others may have upgraded to new schema.20:24
paul_nuclear_eclipse: ? :)20:24
vb1234:)20:24
vb1234paul_: when do you think we can get these fixes done?20:24
vb1234(not sure what they are by the way)20:24
vb1234I just want us to cut the release as agreed earlier.20:25
* paul_ nods20:25
paul_anyway, dates stuff came up mid week20:26
paul_and should be easy20:26
vb1234ok sounds good.20:26
Heady|when the release is planned?20:26
paul_just probably not on a friday night :P20:26
Heady|lol20:26
vb1234are the dates issues release blocking?20:26
vb1234what is the bug #?20:26
paul_apparently version updates on projects feed from a date string straight to db20:27
vb1234nuclear_eclipse: will you get a chance to update docbook with the new plugin events / configuration options?20:27
paul_john needs to look at my new plugin events too20:27
vb1234btw, there is a swekey authentication plugin on the forums (see customization forum).  It might be worth looking at that.20:28
paul_I dont like it20:28
vb1234I haven't looked at the code, but why?20:28
vb1234do we need to add more events?20:28
vb1234We need to use the community plugins as feedback into the events / infrastructure design.20:29
paul_well, we dont support authentication plugins atm (As part of events structure)20:30
nuclear_eclipsevb1234: I'm already working on the docbook for events :)20:30
paul_so it hacks itself in20:30
vb1234agreed, but in this specific example, it is a good use case to consider in our design.20:30
vb1234forget about the specific implementation.20:30
vb1234basically this is the use case where there is a device involved.20:31
paul_I'm also disappointed that they don't appear to release source code for their plugins.20:33
paul_and/or customisations to open source stuff20:33
vb1234interesting.  So you have to buy the key to get the updates?20:33
paul_actually ,tell a lie, they do20:34
paul_I thought that the putty stuff was only the binary20:35
paul_but it does contain source too20:35
nuclear_eclipsevb1234, paul_: bug 1056520:35
mantisbotBug 10565 - jreese - open - assigned20:35
mantisbotUser input date strings are munged with new schema - http://www.mantisbt.org/bugs/view.php?id=1056520:35
paul_this is where it would be easier if we'd commited my bugobject changes branch ;/20:37
paul_$t_version->date_order = $f_date_order;20:37
paul_is that field the date of a version?20:37
nuclear_eclipseyep20:37
paul_(why was it called 'order'20:37
nuclear_eclipseno clue20:37
paul_that makes me thinkg of ASC / DESC20:37
nuclear_eclipsewell, it's the field used to order the versions by :P20:38
mantisbotNew bug: Bug 10565 - jreese - open - assigned20:38
mantisbotNew bug: User input date strings are munged with new schema - http://www.mantisbt.org/bugs/view.php?id=1056520:38
nuclear_eclipsety mantisbot20:38
paul_lol20:38
paul_this is where i'd like type checking on objects :P20:38
nuclear_eclipsepaul_: tbh, I hadn't realized until two days ago that you hadn't merged bug objects into master yet...20:39
paul_tbh20:39
paul_I hadn't realised until 10 minutes ago20:39
nuclear_eclipselol20:39
nuclear_eclipseI'd assumed you had done it at the same time as the date schema..20:39
nuclear_eclipseesp considering you never brought it back up after that20:40
paul_how can i delete local branches20:42
nuclear_eclipse`git branch -d <branch>`20:42
nuclear_eclipsetime to head home; I'll try to be on again in about 30 mins20:55
nuclear_eclipsecheers20:55
* paul_ yawns22:21
nuclear_eclipsesiebrand: why do we need to use `lang_get( 'word_separator' )` that returns `&#32;` that breaks a lot of things rather than just using ' '?22:57
nuclear_eclipseeg, on the roadmap, we use word_separator between the version and the filter link, but the browser doesn't render the space appropriately22:58
nuclear_eclipseok, let me refine that, Chrome doesn't render it correctly22:59
nuclear_eclipsebut still, why use "&#32;" instead of just " "?23:00
siebrandnuclear_eclipse: well, " " is so empty. At least #32 will show it's a space. But there is no issue with making it a space.23:04
nuclear_eclipsethe other reason I ask is because of emails; &#32; does not render correctly in our plaintext emails23:05
nuclear_eclipseit actually ends up as "word&#32;word" in the email23:05
nuclear_eclipsesiebrand: is there any language where word_separator would *not* just be " "+23:06
nuclear_eclipse?23:06
siebrandnuclear_eclipse: well, then it should be rendered to plain text before being put in there?23:06
siebrandnuclear_eclipse: it can be made a 'real' space.23:06
nuclear_eclipsehow?23:06
siebrandnuclear_eclipse: good question. I have no idea. Let me check how MediaWiki does that, if I can find out..23:07
nuclear_eclipselol23:07
siebrandhmm, i think I know how (hard coded transform)23:08
siebrandlet me look it up23:08
siebrandnuclear_eclipse: yep: http://www.mediawiki.org/wiki/Special:Code/MediaWiki/5145323:09
nuclear_eclipseoh, well I could have figured that much out; I was looking more for a *real* method of converting HTML entities to plain text...23:11
siebrandhmm, intersting Google logo today.23:12
nuclear_eclipseah, html_entity_decode()23:12
nuclear_eclipselooks normal to me...23:12
vb1234normal to me!23:12
vb1234Bing!23:12
nuclear_eclipsehi again Victor23:13
vb1234hi nuclear_eclipse23:13
nuclear_eclipsesiebrand: is there any language where word_separator would *not* just be " "?23:14
siebrandnuclear_eclipse: japanese, chinese, some other Asian languages.23:15
siebrandmediawiki has it for lzh, yue, zh-hant and zh-hant at the moment.23:15
siebrandit is an optional message, so few translators actually are aware of it.23:16
nuclear_eclipseok, so would I be safe to convert any lang_*'s word_separator to " " as long as it's already "&#32;", and leave it alone if it's something different?23:17
siebrandnuclear_eclipse: yes. There are only two options: a space or no space.23:18
nuclear_eclipseyou lost me at the "two options" thing23:19
siebrandMediaWiki has the following separators, although using them in general is discouraged (although unavoidable at times):23:19
siebrand'catseparator'        => '|', # only translate this message to other languages if you have to change it23:19
siebrand'semicolon-separator' => ';&#32;', # only translate this message to other languages if you have to change it23:19
siebrand'comma-separator'     => ',&#32;', # only translate this message to other languages if you have to change it23:19
siebrand'colon-separator'     => ':&#32;', # only translate this message to other languages if you have to change it23:20
siebrand'pipe-separator'      => '&#32;&bull;&#32;', # do not translate or duplicate this message to other languages23:20
siebrand'word-separator'      => '&#32;', # only translate this message to other languages if you have to change it23:20
siebrandand then there's the special case:23:21
siebrand'and'           => '&#32;and',23:21
siebrandthat's use in language specific 'running text lists', for example.23:21
siebrand"a, b, c, and d"23:21
paul_nuclear_eclipse: whats the problem with 32/23:21
siebrandpaul_: that is was being put in mails as "&#32;" instead of " ".23:22
nuclear_eclipsea) it doesn't render correctly in Chrome, and b) it renders as "&#32;" in plaintext emails... :P23:22
siebrandoh, chrome has a bug. Interesting.23:22
nuclear_eclipseeither chrome or webkit23:22
nuclear_eclipseI only have Linux, so I can't test safari atm23:22
siebrandnuclear_eclipse: have a sample URL, I'll check with Safari, too.23:22
nuclear_eclipseyeah, look at the mantisbt.org roadmap23:23
nuclear_eclipseon FF3, it shows "1.1.8 [ View Issues ]"23:23
nuclear_eclipseon Chrome, it shows "1.1.8[ View Issues ]"23:23
paul_right23:23
paul_1.1.8</a><span class="bracket-link">[&nbsp;23:24
paul_there's no &#32 in the source?23:24
siebrandusing Windows, I see it wrong in both FF and Chrome.23:25
nuclear_eclipselol23:25
nuclear_eclipseso, even more reason to move back to " " :P23:25
paul_whats wrong with &nbsp?23:25
siebrand>1.1.8</a><span class="bracket-link">[&nb23:26
siebrandexpecting ">1.1.8</a>[space here]<span class="bracket-link">[&nb" right?23:26
nuclear_eclipsepaul_: what's wrong with " "? :P23:27
nuclear_eclipseIMO, it's the "non-breaking" part of "non-breaking space" that I don't like about &nbsp;23:27
siebrandthis is the line of code we are discussing, right?23:28
siebrandecho '<br />', $t_release_title, $t_scheduled_release_date, lang_get( 'word_separator' ), print_bracket_link( 'view_all_set.php?type=1&amp;temporary=y&amp;' . FILTER_PROPERTY_PROJECT_ID . '=' . $t_project_id . '&amp;' . filter_encode_field_and_value( FILTER_PROPERTY_TARGET_VERSION, $t_version_name ), lang_get( 'view_bugs_link' ) ), '<br />';23:28
nuclear_eclipseyep23:28
siebrandso where did that space go? What ate it? It's not in my HTML source.23:29
nuclear_eclipsethat's a fantastic question; FF3 on Linux renders it correctly, but only shows " " in view source23:29
* siebrand looks confused.23:30
nuclear_eclipseI think I'd really prefer to just drop &#32; -- it seems to be doing some really funky things23:30
siebrandnuclear_eclipse: and if you use a plain space it'll be better?23:31
nuclear_eclipseit'll at least be predictable :P23:32
siebrandnuclear_eclipse: you have server access, right?23:32
siebrandcan you replace it in strings_english.txt for a second?23:33
nuclear_eclipsesure23:34
nuclear_eclipsechanged23:36
siebrandno change in output.23:36
siebrandspace still eaten (FF3)23:36
nuclear_eclipsehmm23:37
nuclear_eclipsetry CTRL-F523:37
siebrandnot changed in Chrome either. Did the hard refresh...23:37
siebrandchecking safari on windows on another computer....23:38
nuclear_eclipsehmm, not changed for me either in Chrome, although it fixed it on my local machine...23:39
nuclear_eclipseoh!23:39
siebrandalso no space on Safari23:39
nuclear_eclipseit's APC23:39
siebrandapc is eating the space, or still using &32?23:41
nuclear_eclipseAPC is the opcode cache23:41
siebrandyep, I knwo.23:41
siebrandthat should invalidate on file change, shouldn't it?23:42
nuclear_eclipsehmm, tried clearing its cache, and that didn't fix it23:42
siebrandOK, so whatever it is, there's no space where we expect it. That sounds like a challenge :)23:42
siebrandtry putting a &nbsp; in word_separator (just for testing! :P)23:43
siebrandthat must really appear..23:43
nuclear_eclipsethe *real* oddity is that it looks right in FF3, both before and after the change, and wrong in chrome, both before and after, but *right* in chrome after I make the change on my local test install....23:43
siebrandI don't know how to create roadmap/changelog entries, so I cannot check on my local install.23:44
nuclear_eclipseadd versions to a project, and then target those versions from issues23:45
nuclear_eclipseI've reverted live, just fyi23:47
siebrandThis is my local output:23:49
siebrandproject2 - 1.1 (Scheduled For Release 2009-06-06) [ View Issues ]23:49
siebrandLet's get rid of that date...23:50
siebrandehr: project2 - 1.1 (Scheduled For Release 1970-01-01) [ View Issues ]23:50
* siebrand grins.23:50
siebrandLooks like leaving the scheduled release date empty isn't an option.23:51
siebranddate_order = 0 in db (Mysql)23:52
siebrandIs that expected behaviour?23:52
nuclear_eclipsethat's also currntly a bug in how dates are (not) working atm23:53
siebrandanyway, it looks like "word_separator" works for me in local version.23:53
nuclear_eclipsesame...23:54
siebrandlooks OK in Chrome, too.23:54
nuclear_eclipsethis is really weird, because it wasn't working earlier...23:54
siebrandand our own tracker is git master head, too?23:54
siebrandwell, my preliminary conclusion is that it's not as broken compared to " " as you made it sound :)23:56
nuclear_eclipseah ha!23:56
* siebrand listens carefully.23:56
nuclear_eclipsesiebrand: in your local, do `git checkout live` and test23:56
nuclear_eclipseyou may need to fetch first if you haven't recently23:57
siebranderror: pathspec 'live' did not match any file(s) known to git23:57
nuclear_eclipseyou may need to fetch first if you haven't recently23:57
siebrandI need some other git magic first?23:57
nuclear_eclipse`git fetch`23:57
nuclear_eclipseit's a new public branch representing where the live site is23:57
siebrandstill won't work.23:58
siebrandAnyway, sounds like you can reproduce it.23:58
siebrandSo this was fixed somewhere recently?23:58
nuclear_eclipseI don't know23:58
nuclear_eclipsegit-bisect to the resuce!23:58
nuclear_eclipserescue even23:58

Generated by irclog2html.py