Tuesday, 2009-05-19

../irclogs/#mantishelp.2009-05-19.log
--- scribe started ---00:00
markwanyway to get the file uploads to go bigger than 8M?04:29
markwphp.ini set to 20M04:29
markwand the g_max_file_size or whatever is set to 20m.04:29
markw20M.04:29
mantisbotNew bug: Bug 10497 - Mr Papa - open - new06:22
mantisbotNew bug: default version and target_version - http://www.mantisbt.org/bugs/view.php?id=1049706:22
mantisbotNew bug: Bug 10498 - mstedman - open - new06:27
mantisbotNew bug: Allow search filter to include issues within "disabled" projects and their children - http://www.mantisbt.org/bugs/view.php?id=1049806:27
giallumarkw, which storage method? DB?06:37
mantisbotNew bug: Bug 10499 - UmeshKhairnar - open - new06:42
mantisbotNew bug: Published Exe's are throwing Exception on server - http://www.mantisbt.org/bugs/view.php?id=1049906:42
[KK]Kirillhi all07:55
[KK]Kirillwho use last git?07:56
[KK]Kirillwhat happend with datetime?08:01
[KK]KirillI have date 01.01.197008:02
mantisbotNew bug: Bug 10500 - mantistestor - open - new08:12
mantisbotNew bug: Memo, Textarea, Multiline Customfields - http://www.mantisbt.org/bugs/view.php?id=1050008:12
gialluuhm08:35
[KK]Kirillgiallu: hi08:35
gialludatabase_version suddenly is 17408:35
gialluhi [KK]Kirill08:36
[KK]KirillCan you help me debug bad-date?08:36
gialludid you run the install.php upgrade steps?08:40
[KK]Kirillno :)08:43
[KK]KirillAttempting to connect to database as user  GOOD08:44
[KK]Kirillthat's all08:44
[KK]Kirillsorry08:44
[KK]Kirillthis line last08:44
[KK]Kirillit's ok?08:44
[KK]Kirillnot - it's not ok08:47
[KK]KirillI press upgrade, but not happend08:52
[KK]KirillChecking Database Server Version08:52
[KK]KirillRunning mysql version 5.0.77 GOOD08:52
[KK]KirillInstalling Database08:52
[KK]Kirillthat's all.08:52
[KK]Kirillgiallu: ?09:10
paulr_kirill: what database version is listed in select * from mantis_config_table where value = 'database_version'09:10
[KK]Kirillempty value09:11
[KK]Kirill0 row09:12
[KK]Kirillhi paulr09:12
* paulr_ wonders if he spelt it wrong09:12
paulr_just do select * from mantis_config_Table and find the db version :P09:12
[KK]Kirill8309:13
[KK]Kirillwhere config_id = 'database_version'09:13
paulr_that's out of date then09:15
paulr_install.php?09:15
paulr_in fact 83 is out of date before my changes09:15
paulr_git trunk was on 8609:15
[KK]Kirilltry /admin/install.php09:15
[KK]KirillAttempting to connect to database as user  GOOD09:16
[KK]KirillChecking Database Server Version09:16
[KK]KirillRunning mysql version 5.0.77 GOOD09:16
[KK]KirillInstalling Database09:16
[KK]Kirillthat's all09:16
[KK]Kirillone sec09:17
[KK]KirillI think it's my mistake09:20
[KK]Kirillit's normal SYSTEM WARNING: htmlentities() expects parameter 1 to be string, array given ?09:22
[KK]Kirill*rofl* Your database has not been created yet. Please create the database, then install the tables and data using the information above before proceeding.09:22
[KK]Kirillok.. It's my one more mistake09:25
[KK]KirillYes.. all work09:26
[KK]KirillHow I can correct some bad date?09:27
[KK]Kirillpaulr: Thanks09:35
[KK]Kirillpaulr: How I can use Firephp for debug?09:42
[KK]Kirillor it's only for events witch I write in config_inc.php?09:43
Raging_HogHow can I persuade my colleagues to use mantis instead of trac? What I've found out, it seems like trac doesn't have relationships between issues, ability to change statuses of many issues easily and you have to dig more to get a custom query09:44
Raging_HogBut, trac does import CSV files. I couldn't get mantis to import anything even with wget09:48
paulr_[KK]Kirill: so you've fixed the date stuff?10:07
Raging_Hogversion 1.1.6, I can't put issues on roadmap. I specify a target version, the issue has a priority, is new or assigned and has severity and reproducibility. Clicking on the roadmap-link just shows the empty page like when clicked on empty changelog. Should I do something else?10:15
[KK]Kirillpaulr_: yes10:23
[KK]Kirillpaulr_: thanks10:23
[KK]KirillI copy/past last date value to fields where value equal 110:24
paulr_erm10:33
paulr_did you have to do that manually????10:33
[KK]Kirillyes10:33
[KK]Kirillwith phpmyadmin10:33
paulr_sorry, what was the initial value before you modified?10:33
[KK]Kirill01.01.197010:34
[KK]Kirillafter upgrade 110:34
[KK]Kirillit's because I have renamed admin folder10:34
paulr_so the upgrade set it to 1?10:37
paulr_or what ;/10:37
[KK]Kirillpaulr_: datefield afer upgrade has value 1.10:53
paulr_ok so that's fine10:53
[KK]KirillI change it to value from last correct issue10:53
paulr_1 is correct for fields that haven't been used10:54
[KK]Kirillyes10:54
* paulr_ doesnt' get what needed changing?10:55
[KK]Kirillpaulr_: nothing.. i haven't problem.. for now after upgrade I delete admin folder, before I upload new admin-folder11:04
mantisbotNew bug: Bug 10501 - informatyk - open - new13:54
mantisbotNew bug: It takes about 25seconds to load the login_page.php - http://www.mantisbt.org/bugs/view.php?id=1050113:54
DirtyAlhello14:59
DirtyAlneed some help with the mantis SVN integration14:59
DirtyAlhave done alllisted in http://leetcode.net/blog/2009/01/integrating-git-svn-with-mantisbt/14:59
DirtyAlI just need some  tips on how ti reflect commits done in SVN in mantis15:00
DirtyAlhow to I use the post-commit script15:00
DirtyAltks15:00
paulr_DirtyAl: nuclear_eclipse who wrote that normally lurks around this time of day15:20
DirtyAltks, I think paulr_, lets see if nuclear_eclipse can help me16:00
markw_heh, fell out of the buffer.   Looking at increasing file upload size.  php.ini set to 20M, LimitRequestBody set to 20971520, and storage is on disk.   "report issue" page says 8,000k max attachment.16:35
markw_where is it coming up with the 8,000k limit when I'm trying to allow 20M?16:38
paulr_a) I think there might be a max file size in config.inc.php or whatever16:48
paulr_b) mysql limit?16:48
paulr_bbiab16:48
DirtyAlI need some help on how o use the post-commit script from the SourceWebSVN plugin16:50
DirtyAltks16:50
markw_paulr: I set max_file_size in confi.inc.php to 20971520 and went to disk storage.16:59
markw_ha, found it.  max_post_size.17:04
DirtyAlhas anyone seen this messages?17:28
DirtyAlAPPLICATION WARNING #2400: Event "EVENT_ACCOUNT_PREF_UPDATE_FORM" has not yet been declared.APPLICATION WARNING #2400: Event "EVENT_ACCOUNT_PREF_UPDATE" has not yet been declared.17:28
DirtyAlafter installing the SVN/mantis integration plugins?17:28
DirtyAlI manage the post-commit script17:53
DirtyAlit is updating the changesets on mantis17:54
DirtyAlbut it is not closing the tickets17:54
DirtyAlany help on this matter17:54
DirtyAl?17:54
DirtyAlhello? anyone on the channel?17:56
paulrmoo18:01
DirtyAlhehehe18:05
nuclear_eclipseDirtyAl: you need to be running Mantis from latest Git, or a nightly build18:16
DirtyAlhummm.... tks, I am running 1.2a3, I am seeing the repos, managing them, the tickets dont get automaclty fixed by commit18:18
DirtyAli will download the nighly build right now, tks.18:19
DirtyAlthen I just list the ticket number in the commit message like this #445 and the ticket will be closed and the commit  message attached?18:21
* thraxisp slaps paulr with a large trout18:22
thraxispWhat's up with changing dates to integers?18:23
nuclear_eclipseDirtyAl: you need to enable auto-resolving of issues in the source-integration configuration page before it will do that18:25
nuclear_eclipsethraxisp: it's so that we can actually start handling dates and timezones without dealing with the many stupid ways that DB's handle datetime values...18:26
nuclear_eclipseI'm not sure that I agree with the move to integers, but if it gets paulr to actually *fix* something....18:26
DirtyAlThk nuclear, I already did, but it is no doing so,I guess because I am not using the nighly build.18:28
paulrerm iirc int's was what you / victor agreed on ;/18:28
* paulr slaps nuclear_eclipse thraxisp 18:29
nuclear_eclipseI think most of what you're doing could have been done using the existing datetime schema, which would have made upgrades much more straightforward, but I also agree that using ints instead probably makes the code much simpler; it's always a tradeoff18:30
paulrit certainly speeds up code a little18:30
nuclear_eclipseDirtyAl: also make sure that a) your commit comments are matching the patterns for fixed issues, as opposed to simply linked issues, b) that your mantis username matches your SVN username or that you specify your SVN username in your account prefs18:34
* paulr stares at nuclear_eclipse a bit more18:35
* nuclear_eclipse ignores paulr18:35
paulranyway, our handling of datetime18:36
paulrgoes as much as:18:36
paulrstore datetime in db18:36
paulrretrieve datetime from db as string18:36
paulrconvert string to integer18:36
paulruse integer18:36
nuclear_eclipseDirtyAl: I recommend for the time being that you grab a nightly snapshot built from before the weekend, or grab a snapshot from git on the renamed-libraries-1.2.x tag :P18:37
nuclear_eclipsepaulr: I'm not really sure why we couldn't just continue using that approach for compatibility reasons and to prevent the need for a whole lot of goofy schema-upgrading business....18:38
paulrit was hardly compatible18:39
nuclear_eclipsewell, I meant "compatible" in the sense that it didn't require a gigantic set of schema changes that could potentially break an upgrade...18:39
paulrbug 797418:39
mantisbotBug 7974 - obouillaud - open - acknowledged18:39
mantisbotDate problem on french MSSQL - http://www.mantisbt.org/bugs/view.php?id=797418:39
paulrwell18:40
paulrwe need to decide quickly if we're now changing our mind on approach18:41
DirtyAlnuclear_eclipse: its working like a charm18:45
DirtyAltks for the help18:45
DirtyAlgreat software18:45
nuclear_eclipseDirtyAl: you're welcome, and thank you :)18:45
paulrso what are you actually saying you want to do john ? :P18:46
nuclear_eclipsedunno18:47
paulruseful ;p18:48
paulrbut seriously, one of the 'issues' with datetime in db and supporting oracle/postgres/mssql/the rest18:48
paulris we are passing strings to db18:48
paulrgetting a string back18:48
nuclear_eclipseusing integers for dates is nicer/better; my only real complaint about it is the sheer amount of schema upgrades required to make that change18:48
paulrconverting the string to an int before using18:48
* paulr nods18:48
nuclear_eclipsewhich is why I don't have a really strong opinion either way; they both have advantages and tradeoffs18:49
paulrI think there's still some work to do/review on int's (within mantis)18:49
paulras you know my timezone is +0, which makes it somewhat more difficult to easily test stuff18:50
paulri'm planning to have a review over next day or two to see what's left/wrong if anything18:50
markw_hey, I noticed something when playing with the DISK file attachments.18:53
nuclear_eclipseyou're still welcome to have an account on my server to test in a -500 timezone...18:53
markw_you can't have duplicate filenames in the directory.18:53
markw_so if someone uploads an attachment called logfile.txt it appears that it will try to save it over an existing logfile.txt18:56
markw_for separate bugs.18:56
DirtyAlnuclear_eclipse:mantis sends emails on each update on a ticket19:04
DirtyAlvia SVN, is does't19:04
DirtyAlcoudl this be done?19:04
thraxisppaulr: using ints makes the db impossible to read or debug.19:04
paulrnot really...19:05
paulrmysql you'd just do select from_unixtime() or whatever19:05
thraxispbut using the MySQL Admin tool is now useless.19:07
nuclear_eclipseDirtyAl: potentially, yes; I just haven't gotten the chance to implement something of that sort19:07
thraxispquick. when is 1234592284919:07
nuclear_eclipseduh, thraxisp, that's so obvious, I'm not even going to answer you :P19:08
DirtyAlnuclear_eclipse: tks, for now I will make the commit from SVN to send emails, it will do19:08
thraxisppaulr: I'm sure that every other php based package in the world that does time zones doesn't store times as ints.19:12
paulrI'm assuming you didn't mean 1234592284919:20
paulranyway, //echo $asctime(<timestamp>) in mirc is quite a quick way to answer that question19:20
paulras is select from_unixtime(x)19:20
paulriirc, when I last looked things vary19:21
paulryour thinking just mysql here19:21
paulrthe first thing we spend time doing with the old code is converting a date string to an int19:22
paulrwe then use the int internally19:26
paulrand if we are putting int back into db convert it to a string for the insert19:26
thraxispNot really. I can only think of a few places where an int is needed.19:27
paulrby default we convert it to an int19:27
paulrwithin the api19:27
thraxispBut it's readable in the database as a timestamp.19:27
paulryou need to get the datetime from a string to an int for us to then print it as a date19:28
paulrmaybe i'm missing something here but19:28
paulryou spend hours reading the db? :)19:28
thraxispYou are making it impossible to find a bug created yesterday in a database dump or via the db admin tools.19:28
thraxispI have when debugging an issue.19:28
paulrwell, that depends on your db admin tool ofc...19:29
thraxispI also have used other packages that store dates as ints and waste hours converting to see what actually happened.19:29
* paulr nods19:29
thraxispThe MyQSL tools cone to mind19:29
thraxisp^cone^come19:29
paulrbut even so our standard behaviour19:30
paulrbefore this change was19:30
paulrwhenever we load a date from db, convert it to an int for use in php19:30
thraxispThat is fine because it's hidden. There is no reason to corrupt the database because of a 0.01% performance improvement.19:31
paulrfor example: http://git.mantisbt.org/?p=mantisbt.git;a=blob;f=core/history_api.php;h=c7f36edbea4de58aa3977d7ca3d7d3cec91f1b2e;hb=master-1.1.x19:31
paulrfunction history_get_raw_events_array( $p_bug_id, $p_user_id = null ) {19:31
paulr$raw_history[$j]['date']        = db_unixtimestamp( $v_date_modified );19:31
paulrfor each result19:31
paulrbut i'm wondering a bit here...19:32
paulrlets say we reverted this commit19:32
paulrif I have a mysql server in germany atm19:33
paulrand I want to move it to the UK (different timezone)19:33
paulratm, in mysql with datetime, we store dates as UTC?19:33
paulrMy reasoning for changing was a few issues19:34
thraxispPersonally, I'd read the server timezone from the host, and have a user timezone. I would use the difference for display purposes. The database would be the server time.19:35
paulra) we've got some bugs against mssql that you can configure mssql to handle dates as either MDY or DMY (whichever it should be)19:35
thraxispThis assumes that moving the server is rare. Even so, it takes a few SQL commands to update all of the timestamps.19:35
paulrI'm not quite sure how we detect in php what collation dates are stored in in db all the time19:35
paulrb) duedate's stuff is broken19:36
paulras the calendar code is using utc/gmt/whatever19:36
paulrthe database code is using something else19:36
paulrand php/adodb are trying to convert it to an int19:36
paulrand we're trying to compare int's at the end19:36
paulrc) regarding the database19:37
paulrsurely, if you were going to pull a timestamp from db19:37
paulrand localise it19:37
paulryou'd need to have db returning stuff without a timedifference applied19:37
paulr(or at least parsing it like that)19:37
paulrI dont mind that much whether we store dates or int's in db19:38
paulrbut trying to worko ut how mssql/pgsql/oracle/whatever else people ask about19:38
paulrhandle dates19:38
paulrand the fact that the adodb library generates a warning in it's date routines for every call (As it calls mktime() without a param)19:38
paulractually was starting to add up to more of a performance issue19:39
paulrI'm kinda assuming that it's not an issue to store int's in a db, and gives us a slight performance increase, and should remove some headaches19:41
paulron the basis that:19:41
paulra) by default when we select a datetime from db, the first thing we *always* seem to do is to convert it to an int (summary api does that a few thousand times)19:41
paulrb) it's rare to look at a database in raw mysql format for a webapp - as if the web app is working, you wouldn't need to, if your debugging something, your likely to look at the php19:42
ivan_htphey guys19:51
ivan_htpi'm trying to integrate dokuwiki with mantis, but the single sign on doesnt works.. could anybody help me?19:52
thraxispYou seem to have a penchant for optimizing the wrong things. Saving 1 or 2 C function calls won't make any real difference in real time.19:54
thraxispI'm sure we're going to spend the next 6 months fixing everything that relates to dates.19:54
paulrthe point of changing dates was to try to fix what's broken atm19:55
thraxispbut you need to think of all of the other things that you did/might break in the process.19:55
* paulr nods19:55
paulrthat's why the branc has been around for a month or two19:56
thraxispI have a few migration scripts that translate from other trackers. These are all toast now.19:56
thraxispI assume you are going to fix them too.19:56
paulrI'm happy to fix anything i've broken.19:56
paulrsimilarly, i'm happy to remove the code if needbe19:57
paulrbut we've been talking for ages on how to handle dates19:57
thraxispBut have you tested it fully? (The debian test plan comes to mind).20:03
paulrdhx/myself have - i'm planning to review/test it more this week20:04
paulrit's a shame you didn't pick up on the discussion(s) that lead to this patch20:06
paulrthraxisp: if you take the (rather biased) blog post at http://billauer.co.il/blog/2009/03/mysql-datetime-epoch-unix-time/20:09
paulrthey do:20:09
paulrSELECT * FROM stupid_date;20:09
paulrto get:20:09
paulr 2009-03-15 14:01:4320:09
paulrmy issue was:20:09
paulr(iirc at least)20:09
paulryou dont know the timestamp on that20:10
thraxispI'm unhappy that the database is no longer readable. W should ROT13 the data as well. It just makes it hard to debug / fix anything.20:10
paulrI've noticed20:10
paulrbut surely it's not that much harder to do:20:11
paulrSELECT FROM_UNIXTIME(last_updated) from mantis_bug_table20:11
paulron the occassion that one needs to read the db20:11
paulrI think one of the issues here is20:11
paulryou use some gui?20:11
paulrwhereas, in my case at least, I tend to just use the cmd line mysql command20:12
thraxispThe author has obviously never been a DBA.20:12
paulrso if i'm typing a query, adding 'from_unixtime' isn't that big a job20:12
thraxispNo, byt generally, it's select * from table where foo=1;20:13
paulras I said it's a rather biased view :)20:13
thraxispThe date is now garbage.20:13
paulrnod, unless one explicity from time's it20:14
paulrbut then, i'm looking at issues like http://www.mantisbt.org/bugs/view.php?id=797420:15
paulrand due dates20:15
paulrpart of the reasoning of converting to ints was to get rid of all the crap we've got going on to try to handle converting an int to a string to put it in a db20:17
paulrI'm pretty sure at one point of looking at fixing duedates I had a scenario that went along lines of:20:18
paulrjava script calendar has date picker on20:18
paulrreturns a string to php20:18
paulrphp converts string to an int20:18
paulrdatabase api converts into a string20:19
paulr<stored to db>20:19
paulr<retrieve from db>20:19
paulrconvert string to int20:19
paulrconvert int to string for jscript calendar20:19
paulrget string back from jscript calendar20:19
paulrand your 2 hours out20:19
paulr(Which just happens to be your timezone offset from gmt)20:20
paulrnow, put that into the mix with user timezones that are seperate to database20:20
paulrI'm quite confident I can fix any things i've broken converting to ints in less time then it will to figure out that mess :)20:21
thraxispWe need to convert this stuff to OOP and hide the int date.20:30
paulrDateTime::createFromFormat( <int>, 'U' );20:33
paulris probably a sensible 'format' to throw around20:33
paulrs/format/object/20:33
paulranyway, part of my logic is I want to reduce reliance on adodb20:34
paulr(which was handling dates)20:34
paulrI appreciate it's a bit of a pain20:35
paulrI guess it also introduces the 2038 problem20:35
paulrbut I think it should make some things easier20:35
thraxispIs 2038 our next release?20:35
thraxisp:)20:35
paulrstrings in different formats in different db's = headache20:35
paulrint's = well, we can be fairly sure what an int is :)20:35
paulrbut when you've got users going I use mssql in french, and I need to run this custom query or modify DMY in adodb source to MDY to make it work20:36
paulrdoes strike me as hassle20:37
paulr(as i'm pretty sure if we changed DMY->MDY you'd break whichever half of the world was working before :)20:37
thraxispI think you are trading off hours of admin time for a few ms or computer time.20:38
thraxispWe should read and write db date/time and convert to ints internally.20:39
thraxisp^or^of20:39
thraxispadodb should handle the differences between the databases. If it doesn't we should throw it out and use something better (PDO, Doctrine).20:40
paulrpdo doens't handle dates20:44
paulr  * Obtain DBMS specific SQL code portion needed to declare an text type20:48
paulr     * field to be used in statements like CREATE TABLE.20:48
paulr     *20:48
paulrpublic function getNativeDeclaration($field)20:48
paulr            case 'date':20:49
paulr                return 'CHAR(' . strlen('YYYY-MM-DD') . ')';20:49
paulr            case 'time':20:49
paulr                return 'CHAR(' . strlen('HH:MM:SS') . ')';20:49
paulr            case 'timestamp':20:49
paulr                return 'CHAR(' . strlen('YYYY-MM-DD HH:MM:SS') . ')';20:49
paulrso it seems that doctrine users char() for dates in mssql20:49
paulrdate/datetime types in mysql20:49
paulri.e. in short, if/when I find something better then adodb20:51
paulrI'll be the first to want to throw it out :)20:51
paulrthe closest i've come so far is actually http://ezcomponents.org/20:52
thraxispI would worry that by the time we find something "better" we have forgotten how the code works.20:56
thraxispThe code, in many places, is obscure at best now. This just confuses it more.20:57
paulrthe current stuff should make it less confusing gradually20:58
thraxispexcept that dates aren't dates in the database.21:17
mib_xd145vhey - i wonder if anyone is doing things like needing to associate a client with a certain # of bugs for a given product21:33
mib_xd145vlike a client is asking for features in your core application and you want to group by the clients requess... i imagine at this point you could use the wiki to mention the bug id's but any other ideas?21:34
mantisbotNew bug: Bug 10502 - mclaus - open - new21:36
mantisbotNew bug: Vraag om te doen - http://www.mantisbt.org/bugs/view.php?id=1050221:36
thraxispmib_xd145v: You could use a custom field to hold the client name if there is more than one reporter associated with the client. Or you can query by reporter, if there is only one user.21:45
mib_xd145vthraxisp: thx - I have done that in my existing 1.06 version but i think i stopped using it because i was concerned about the drop down getting a bit crazy on that field, but yeah, product with that field would do it wouldnt it.  show only there items and still allow for the nice roadmap funciton for a product overall.. this is great software btw :)22:04
mib_xd145vi have been considering installing the newer stable build because of the roadmap aspect and wiki integration sounds nice as well.22:05
mib_xd145vi am a bit worried since 1.06 seems so stable - has the software become a little shaky with the new features or is the latest stable build a good bet?22:06
paulrwhat version are you thinking about?22:07
mantisbotNew bug: Bug 10503 - misu - open - new22:11
mantisbotNew bug: APPLICATION ERROR 0001303 when custom float field is left blank - http://www.mantisbt.org/bugs/view.php?id=1050322:11

Generated by irclog2html.py