| ../irclogs/#mantishelp.2009-06-09.log | ||
| --- scribe started --- | 00:00 | |
| mantisbot | New bug: Bug 10572 - sdurkin - open - new | 04:37 |
|---|---|---|
| mantisbot | New bug: APPLICATION ERROR #1802: Relationship not found. - http://www.mantisbt.org/bugs/view.php?id=10572 | 04:37 |
| mantisbot | New bug: Bug 10573 - nobswolf - open - new | 07:43 |
| mantisbot | New bug: support SQLite - http://www.mantisbt.org/bugs/view.php?id=10573 | 07:43 |
| giallu | SQLite rocks | 07:46 |
| CIA-61 | Mantisbt: s.mazeland * rd9978e58107f /lang/strings_english.txt: Consistent capitalisation | 09:36 |
| nuclear_eclipse | morning all | 12:27 |
| Kirill_Krasnov | nuclear_eclipse: lo | 12:28 |
| [KK]Kirill | John, I have wish | 12:29 |
| nuclear_eclipse | uh oh | 12:29 |
| [KK]Kirill | :) | 12:29 |
| [KK]Kirill | I need 2 events | 12:29 |
| [KK]Kirill | before create/update user and after update create user | 12:29 |
| nuclear_eclipse | so, four events? :P | 12:30 |
| [KK]Kirill | 6 | 12:30 |
| [KK]Kirill | and delete | 12:30 |
| [KK]Kirill | before after | 12:30 |
| nuclear_eclipse | why both befre and after delete? | 12:30 |
| nuclear_eclipse | I should think a single delete event would suffice | 12:31 |
| [KK]Kirill | no.. 2 events before and after delete | 12:31 |
| [KK]Kirill | if user can't be deleted? | 12:31 |
| [KK]Kirill | or I don't want delete him in plugin | 12:31 |
| nuclear_eclipse | if there's an event before the deletion, I don't see how the second event adds any other ability | 12:31 |
| [KK]Kirill | for example I check user and reject this operation | 12:32 |
| nuclear_eclipse | why would a plugin reject deletion of a user? | 12:32 |
| [KK]Kirill | I can't write example now, but I think I found for why | 12:34 |
| [KK]Kirill | for now I want any manipulation with users in other system. | 12:37 |
| [KK]Kirill | ex. dokuwiki, smf | 12:37 |
| dhx_m | hi | 12:41 |
| [KK]Kirill | dhx_m: lo | 12:41 |
| nuclear_eclipse | hi dhx_m, long time no see :) | 12:41 |
| dhx_m | would the before delete event take place before Mantis checks if the user is authorised to delete the account? | 12:41 |
| dhx_m | :) | 12:41 |
| dhx_m | I've been watching development here still | 12:41 |
| [KK]Kirill | dhx_m: no. it must be after | 12:42 |
| nuclear_eclipse | dhx_m: I'd assume the event would take place *right* before the deletion... | 12:42 |
| dhx_m | so in that case, why do you need an after event (if you already know for sure the user will be deleted)? | 12:42 |
| nuclear_eclipse | that was my point :) | 12:43 |
| dhx_m | yep | 12:43 |
| [KK]Kirill | dhx_m: ok | 12:43 |
| dhx_m | actually I know | 12:43 |
| [KK]Kirill | nuclear_eclipse: ok | 12:43 |
| [KK]Kirill | I understand | 12:43 |
| [KK]Kirill | 5 events | 12:43 |
| [KK]Kirill | :) | 12:43 |
| dhx_m | if the user is deleted from the database... and a plugin needs to remove data from another db table after the user is deleted | 12:44 |
| nuclear_eclipse | I'm also not fond of allowing plugins to reject user deletions | 12:44 |
| dhx_m | that action may need to happen after the user is removed from mantis_user_table | 12:44 |
| [KK]Kirill | dhx_m: no | 12:44 |
| [KK]Kirill | before | 12:44 |
| [KK]Kirill | Ok. simple example | 12:45 |
| nuclear_eclipse | eg, allowing plugins to reject deletions *does* make a second event necessary, but I don't like that idea; mantis should make the decision, and just send a single event before deleting to allow plugins to clean up any datasets necessary, and then delete the user | 12:45 |
| dhx_m | yep I would normally agree | 12:45 |
| [KK]Kirill | one sec.. I just see event sytem | 12:46 |
| dhx_m | I can't see a current need for an after delete event, but it is something that can easily be added in the future if required | 12:47 |
| [KK]Kirill | function event_signal( $p_name, $p_params = null, $p_params_dynamic = null, $p_type = null ) { | 12:48 |
| [KK]Kirill | we have $p_params | 12:48 |
| [KK]Kirill | what's about return reject in params? | 12:48 |
| nuclear_eclipse | [KK]Kirill: I don't understand why a plugin should be able to reject deleting a user | 12:49 |
| nuclear_eclipse | that just doesn't make sense to me | 12:49 |
| [KK]Kirill | simple example, I check able delete user in other systems | 12:50 |
| [KK]Kirill | if I can't delete in other - I reject delete in mantis | 12:50 |
| [KK]Kirill | It very simple example in my mind | 12:51 |
| nuclear_eclipse | [KK]Kirill: I think that just breaks user interaction and expectations | 12:51 |
| nuclear_eclipse | not to mention, it overcomplicates the subject | 12:51 |
| [KK]Kirill | brr, sorry I don't understand | 12:51 |
| dhx_m | can I ask why we're including a complete copy of 3rd party dependencies within the main mantis repository | 12:51 |
| dhx_m | rather than create a separate repository (local clone) that includes Mantis patches? | 12:52 |
| nuclear_eclipse | dhx_m: adodb and phpmailer you mean? | 12:52 |
| dhx_m | yep | 12:52 |
| dhx_m | seems odd to me that they're bundled with source code for mantis | 12:52 |
| [KK]Kirill | ok | 12:52 |
| dhx_m | I guess those libraries don't use SCM which makes it harder (or they use weird stuff) | 12:52 |
| nuclear_eclipse | a) that's how we've always done it, b) it makes mantis installs much much simpler, and c) upstream for both those libs sucks | 12:53 |
| [KK]Kirill | nuclear_eclipse: I translate you text | 12:53 |
| [KK]Kirill | nuclear_eclipse: may be | 12:54 |
| [KK]Kirill | nuclear_eclipse: What can you offer? | 12:54 |
| dhx_m | nuclear_eclipse: true, and I guess most package managers don't really understand web scripts | 12:55 |
| dhx_m | [KK]Kirill: I agree that a plugin may want to prevent a user from being deleted, but are there any plugins (or plans for plugins) that would need to do it right now? | 12:56 |
| nuclear_eclipse | dhx_m: our Fedora and Debian maintainers both do what they can to separate out the libs when they generate packages for their distros, but it also means issues because upstream doesn't seem to ever pay attention to our patches.... | 12:57 |
| dhx_m | I'm not sure what the process is for adding events, whether they're added now (before a use is found) or whether they're added on demand | 12:57 |
| nuclear_eclipse | dhx_m: kind of both | 12:57 |
| [KK]Kirill | nuclear_eclipse: update channel topic | 12:57 |
| nuclear_eclipse | if it's something that I can anticipate a need for, I tend to add them as soon as I can understand their purpose, and either write a throwaway plugin to test it, or write a full plugin to test it | 12:58 |
| dhx_m | nuclear_eclipse: yep, not much you can about their method of packaging and ignoring patches (they do have valid reasons for using common upstream libs in place of bundled copies) | 13:00 |
| nuclear_eclipse | and I certainly don't blame them | 13:00 |
| dhx_m | nuclear_eclipse: OpenOffice was/is notorious for bundling copies of quite a few libraries and that is generally a bad thing | 13:00 |
| nuclear_eclipse | I do appreciate all the work that aptituz_ and giallu go through; I know there are a lot of people that run Mantis from their packages | 13:01 |
| * giallu is around | 13:01 | |
| dhx_m | nuclear_eclipse: for example, updating a library because of a security flaw... you don't want to wait for every application on your system to release a new version (complete with updated bundled libraries) | 13:01 |
| nuclear_eclipse | the only reason I don't use packages for webapps is because distros never stay on the bleeding edge :P | 13:01 |
| nuclear_eclipse | hi giallu | 13:01 |
| dhx_m | yep, they're very behind on web scripts | 13:01 |
| dhx_m | and many distros just aren't setup properly to handle them nicely | 13:02 |
| nuclear_eclipse | dhx_m: with a lib security issue, you don't want to *find* all the packages bundling that lib to make sure they're up to date... | 13:02 |
| giallu | nuclear_eclipse, solution is to use Fedora ;) | 13:02 |
| nuclear_eclipse | yeah, but that would mean I have to use *Fedora*; don't make me vomit ;) | 13:03 |
| dhx_m | yep that is what I was saying :) | 13:03 |
| dhx_m | I guess the common scenario is creating a symbolic link from the wwwroot to the Mantis scripts in /usr/share/? or some other common directory | 13:03 |
| giallu | if you're not vomiting sorrounded by all those broen stinky stuff, you'll be fine in the blue | 13:03 |
| dhx_m | and how you go about changing config.php type files I'm not 100% sure | 13:03 |
| giallu | s/broen/brown | 13:04 |
| nuclear_eclipse | giallu: orange is my favorite color ;) | 13:04 |
| dhx_m | how does debian/fedora/etc handle web scripts? can you do slotted installs like Gentoo where multiple versions can be installed on your disk at once? | 13:04 |
| giallu | dhx_m, config_inc is a symlink to /etc/mantis/config_inc.php | 13:04 |
| dhx_m | giallu: that isn't so good for shared web hosting though | 13:05 |
| giallu | dhx_m, no | 13:05 |
| giallu | or better, that's another topic | 13:05 |
| giallu | which is not covered by my packagin | 13:05 |
| dhx_m | giallu: well the problem I see is what happens if multiple users on the system want to use Mantis on their "web space" | 13:05 |
| dhx_m | giallu: they'll want to update the version of the script independently of each other, and have their own config | 13:06 |
| giallu | dhx_m, not that I'm going to work on that anytime soon, but no one reported a bug in bugzilla for the mantis component, so I assume that's not a common issue | 13:07 |
| dhx_m | giallu: sounds more like people ignore the packaged version of a script completely? | 13:08 |
| giallu | how can I know? | 13:08 |
| giallu | I know that I have some users of the mantis rpm | 13:09 |
| dhx_m | you can't :) | 13:09 |
| dhx_m | yep I know, I'm not having a go at the Mantis RPMs | 13:09 |
| dhx_m | because it is a general problem that affects all web script packages IMO | 13:09 |
| nuclear_eclipse | dhx_m: that just points out yet another reason why SquirrelMail is the only webapp that I ever use from distro packages :P | 13:12 |
| nuclear_eclipse | ugh, meeting time | 13:13 |
| dhx_m | have fun :) | 13:14 |
| dhx_m | paul_: good work with the WIP on your graph branch :D | 13:32 |
| dhx_m | paul_: although I certainly don't like graph_get_font() in graph_api.php (in your new plugin directory) having a hardcoded path to c:\windows\fonts\arial.ttf :) | 13:34 |
| dhx_m | paul_: I was half thinking of trying to get Mantis using gnuplot instead of jpgraph | 13:37 |
| giallu | talking about graphs | 13:39 |
| giallu | I was asked if interested to package phplot.sf.net | 13:40 |
| dhx_m | paul_: the reason I don't like jpgraph currently is the huge amount of memory it requires to generate large-ish graphs | 13:40 |
| giallu | I denied, but can reconsider if we think we finally want to move away from jpgraph | 13:40 |
| dhx_m | jpgraph looks better in terms of features from what I can see | 13:41 |
| giallu | pity is not free... | 13:42 |
| dhx_m | oh yeah you're right... jpgraph is fairly useless then | 13:42 |
| dhx_m | given it can't be used in corporate settings (for profit) | 13:42 |
| * giallu never seen a single graph from mantis | 13:43 | |
| dhx_m | for simple bar graphs, something like phplot might be OK | 13:43 |
| dhx_m | I use graphs regularly with Mantis | 13:43 |
| dhx_m | although there is a lot of work to be done with it | 13:43 |
| dhx_m | that is something I don't mind working on | 13:43 |
| dhx_m | I already tried to make some gnuplot scripts to show Mantis metrics | 13:43 |
| dhx_m | ie. area graph over time showing the number of bugs at each status level | 13:44 |
| dhx_m | what I'm not entirely sure of is how to properly add graph capabilities into Mantis | 13:44 |
| dhx_m | particularly in a way which plugins can provide their own graphs/statistics | 13:44 |
| dhx_m | for instance, creating a graph showing the number of commits per day | 13:45 |
| dhx_m | giallu: it seems paul was trying out ezGraph (sample graphs at http://ezcomponents.org/docs/api/trunk/Graph_gallery.html) | 13:51 |
| dhx_m | it looks quite pretty :) | 13:51 |
| giallu | that's fine ( already packaged in Fedora :) ) | 13:52 |
| dhx_m | I haven't looked much at it, but it seems fairly heavyweight (you get a lot of other unwanted stuff bundled in) | 13:53 |
| dhx_m | ie. you download all the components at once, even though you only want the graph part | 13:53 |
| giallu | packagers usually split it | 13:55 |
| RonaldH | Hi, Looking for dynamic allocation of custom field from database source. I believe this currently will work through a custom function definition in a custom field? | 14:35 |
| RonaldH | Next would be to chain the selected value to another field... this is what I really cannot find wether its possible though custom functions or plugins (1.2 alpha) ? | 14:36 |
| jleb | Hi all. Working with source-control-integration plugin and having some issues trying to connect it to local server Subversion repository. Anybody familiar with how to config that? | 15:45 |
| nuclear_eclipse | jleb: what's the issue? | 15:53 |
| jleb | subversion is running only a SVN connection, not WebDAV or anything. how do i configure the plugin to pick that up? | 15:54 |
| nuclear_eclipse | should be just a matter of entering an SVN-style url hen configuring the repository in Mantis | 15:55 |
| nuclear_eclipse | when* | 15:55 |
| jleb | which type of repo? Generic? | 15:55 |
| nuclear_eclipse | WebSVN | 15:55 |
| nuclear_eclipse | the "generic" type does nothing but display data already in the system; you should always use a specific repository type | 15:56 |
| nuclear_eclipse | jleb: I have to go to lunch; I'll be back in about 45 minutes | 15:57 |
| jleb | ok, i see. that would be why i was having so much trouble. i will try to work from here and check back in later if i'm still stuck | 15:57 |
| jleb | thanks | 15:57 |
| rioch | should all the .php files be executable on the server? I'm getting 404 errors. | 16:34 |
| rioch | well, I think it's 404...it's this: The requested URL /bugs/admin/install.php was not found on this server. | 16:34 |
| nuclear_eclipse | rioch: no, they don't need to be executable | 16:47 |
| nuclear_eclipse | however, you do need to make sure that the webserver process has read access to the files | 16:48 |
| rioch | nuclear_eclipse: hmmm I think my problem is to do with it being on a subdomain of my site. Any chance you can help? | 16:48 |
| nuclear_eclipse | what server are you running? | 16:48 |
| rioch | nuclear_eclipse: yes, it has read access | 16:48 |
| rioch | apache | 16:48 |
| nuclear_eclipse | using vhosts? | 16:48 |
| rioch | hmmm not sure. how can I find that out? | 16:49 |
| nuclear_eclipse | that would be part of how your apache configuration is set up | 16:49 |
| nuclear_eclipse | you'll need to make sure that mantis is contained within either your apache's DocumentRoot, or the DocumentRoot of an apache vhost | 16:50 |
| rioch | ah I think this isn't an issue related to mantis... | 16:51 |
| nuclear_eclipse | probably not | 16:51 |
| rioch | I have /subdomains/httdocs/bugs and /subdomains/httdocs/test. The test one was created when I setup the domain in plesk, and when I put that as my address in a browser I get "no permissions to view the directory"... | 16:52 |
| rioch | ...however, with /bugs it just says it can't find it | 16:52 |
| nuclear_eclipse | hmm, it might also be a plesk issue,which I've never dealt with -- I don't think it's a mantis issue though -- so I'm sorry, but I'm not sure that I'm gonna be able to help | 16:54 |
| nuclear_eclipse | you should probably try talking to plesk/whoever hosts/admins/configured your system | 16:54 |
| nuclear_eclipse | hi Victor | 16:55 |
| rioch | nuclear_eclipse: ok no problem, thanks anyway | 16:57 |
| [KK]Kirill | John, you update avatar? :) | 17:03 |
| nuclear_eclipse | yeah, a couple weeks ago :P | 17:03 |
| [KK]Kirill | nice | 17:04 |
| [KK]Kirill | I just look it in bugtracker | 17:04 |
| [KK]Kirill | Can you assign this issue to siebrand bug 10553 | 17:05 |
| mantisbot | Bug 10553 - datacompboy - open - new | 17:05 |
| mantisbot | Mantis on www.mantisbt.org/bugs/ sends mails in wrong encoding - http://www.mantisbt.org/bugs/view.php?id=10553 | 17:05 |
| nuclear_eclipse | [KK]Kirill: it seems to me that's a duplicate of 10136? | 17:07 |
| [KK]Kirill | not | 17:07 |
| [KK]Kirill | This task about bad header | 17:07 |
| [KK]Kirill | but this issue about wrong email from mantisbt.org | 17:08 |
| [KK]Kirill | I set english only for recieve correct email | 17:08 |
| nuclear_eclipse | 10553 is about emails not using proper encoding, because mantisbt.org is not yet running latest git ... | 17:08 |
| nuclear_eclipse | and 10136 is where the problem was fixed.. | 17:09 |
| [KK]Kirill | :( | 17:09 |
| [KK]Kirill | when it's fixed in mantisbt ? | 17:09 |
| nuclear_eclipse | a) we could cherry-pick the fix to live, or b) wait until we're sure the latest date issues are resolved, and update live to latest master | 17:10 |
| nuclear_eclipse | hmm | 17:10 |
| [KK]Kirill | nice | 17:10 |
| nuclear_eclipse | wait, live *should* have the fix; the fix was committed to master in Feb, but live is running master from Apr... | 17:11 |
| [KK]Kirill | not | 17:12 |
| [KK]Kirill | paul fixed in april | 17:12 |
| [KK]Kirill | 13 | 17:12 |
| [KK]Kirill | look history of issue | 17:12 |
| nuclear_eclipse | he marked it as resolved in April, but if you look at the timestamp on the changeset, it says 2009-02-28 16:21:35 | 17:13 |
| nuclear_eclipse | the fix was committed to git on Feb | 17:13 |
| nuclear_eclipse | 13:13 < nuclear_eclipse> he marked it as resolved in April, but if you look at the timestamp on the changeset, it says 2009-02-28 | 17:16 |
| nuclear_eclipse | 16:21:35 | 17:16 |
| nuclear_eclipse | 13:13 < nuclear_eclipse> the fix was committed to git on Feb | 17:16 |
| [KK]Kirill | ok. may be | 17:16 |
| [KK]Kirill | if siebrand want - I can test this issue together | 17:19 |
| siebrand | [KK]Kirill: as long as I do not have a test system with e-mail, I cannot test it.. | 17:24 |
| paul_ | . | 17:36 |
| paul_ | dhx_m: WIP? | 17:38 |
| paul_ | dhx_m: oh, work in progress | 17:38 |
| paul_ | dhx_m: obviously graph_Get_font is not going to stay as is | 17:38 |
| paul_ | i think phplot might fail | 17:40 |
| [KK]Kirill | paul_: hi | 17:45 |
| paul_ | lo | 17:47 |
| nuclear_eclipse | paul_: bug 10573 | 17:54 |
| mantisbot | Bug 10573 - nobswolf - open - new | 17:54 |
| mantisbot | support SQLite - http://www.mantisbt.org/bugs/view.php?id=10573 | 17:54 |
| nuclear_eclipse | is that possible with adodb? | 17:54 |
| paul_ | I saw that earlier | 17:54 |
| [KK]Kirill | siebrand: I can test this. Did you have own mantis? | 17:55 |
| [KK]Kirill | paul_: and bug 10369 | 17:56 |
| mantisbot | Bug 10369 - Kirill - open - assigned | 17:56 |
| mantisbot | Verification link not work - http://www.mantisbt.org/bugs/view.php?id=10369 | 17:56 |
| sshirley | Hi everyone. I am migrating from Mantis 1.1.0 to 1.1.7 and I have a slight issue. I brought over all my config files (config_defaults_inc, constant_inc, custom_strings_inc). But when I am on my_view_page.php, I cannot see the priority type but instead I see a strange little down arrow. Any clue? | 18:41 |
| nuclear_eclipse | sshirley: that's intended behavior | 18:41 |
| nuclear_eclipse | the down arrow means it's lower-than-normal -- up arrows mean higher-than-normal | 18:42 |
| sshirley | Hmmm | 18:43 |
| sshirley | Not making much sense there. | 18:43 |
| sshirley | Lower-than-normal? Normal what? Ah....priority. | 18:44 |
| nuclear_eclipse | right, it's just a graphical representation of the priority field | 18:45 |
| sshirley | That's odd. Is there a setting somewhere that says 'use graphic or text'? | 18:47 |
| nuclear_eclipse | no, afaik it was a bug in 1.1.0 that caused the graphics to not show up; I'm pretty sure that it's been that way for as long as I've been around Mantis | 18:48 |
| sshirley | Really? Wow. It doesn't happen in 1.1.0 for me. Just 1.1.7. | 18:49 |
| nuclear_eclipse | "afaik it was a bug in 1.1.0" | 18:49 |
| sshirley | Ah | 18:50 |
| sshirley | :-) | 18:50 |
| sshirley | Ahhh. You mean I *should* see the graphic in 1.1.0. Gotcha. Now it seems that I see the graphic in 1.1.7 but not the text. | 18:51 |
| nuclear_eclipse | however, a glance at config_defaults_inc.php turns up "$g_show_priority_text = OFF;" | 18:51 |
| [KK]Kirill | good night | 18:51 |
| sshirley | lol | 18:52 |
| sshirley | You, Sir, are awesome. Once I get this project up and running, I plan on going more in-depth with the options and settings such as that. | 18:52 |
| nuclear_eclipse | btw, 1.1.7 is officially out of date :P | 18:53 |
| sshirley | Damn! That was REALLY recent, wasn't it? | 18:54 |
| rioch | ok, my install is progressing. I get the possible problem that php.ini has register_globals set to ON. I've checked my server, and this is true, but I don't have access to php.ini......is there another way to disable it? | 18:55 |
| nuclear_eclipse | sshirley: yesterday, not even 24 hours ago :P | 18:56 |
| paul_ | heh | 18:57 |
| paul_ | 1.2beta/rc soon too | 18:57 |
| nuclear_eclipse | rioch: you'll need to talk with your hosting about that; it's possible to override php.ini in a .htacess, but they need to allow that | 18:57 |
| rioch | nuclear_eclipse: is it a problem with it enabled? | 18:58 |
| nuclear_eclipse | rioch: I'm not sure offhand, but register_globals is well-known as a security hole for PHP applications | 18:58 |
| rioch | aha ok, thanks. | 18:59 |
| rioch | nuclear_eclipse: in the meantime, what is the admin default username/pwd? | 19:01 |
| vb123 | administrator/root | 19:03 |
| paul_ | lo vb123 ! | 19:04 |
| nuclear_eclipse | hi Victor | 19:04 |
| vb123 | hi paul_ / nuclear_eclipse | 19:04 |
| paul_ | nuclear_eclipse: are you serious about sqlite btw? ;p | 19:04 |
| nuclear_eclipse | actually, yes | 19:04 |
| vb123 | you mean supporting it? | 19:04 |
| nuclear_eclipse | it would at least make testing supremely easy | 19:05 |
| nuclear_eclipse | vb123: bug 10573 | 19:05 |
| mantisbot | Bug 10573 - nobswolf - open - new | 19:05 |
| mantisbot | support SQLite - http://www.mantisbt.org/bugs/view.php?id=10573 | 19:05 |
| paul_ | nuclear_eclipse: are you sure your serious? | 19:06 |
| paul_ | :( | 19:06 |
| vb123 | It would be a could candidate for a newer version of InstantMantis. | 19:06 |
| nuclear_eclipse | with sqlite, we could build a useful test database that we could pass around as a single file for the ability to jump start test installations... | 19:06 |
| nuclear_eclipse | paul_: of course, that's all reliant on adodb supporting sqlite | 19:06 |
| vb123 | agreed. We can also support it only for demo / experimental. | 19:07 |
| vb123 | but people will use it in production anways :) | 19:07 |
| nuclear_eclipse | a quick look at adodb's site shows that they support it, but I don't know enough about what version we use / they support | 19:07 |
| paul_ | nuclear_eclipse: you forgot rule 1 | 19:07 |
| nuclear_eclipse | ironically, sqlite supports better ACID properties than MySQL... | 19:08 |
| vb123 | The best thing would be to support Gears for MantisBT> | 19:08 |
| vb123 | offline support :) | 19:08 |
| nuclear_eclipse | that would require a lot more effort than sqlite ;) | 19:08 |
| vb123 | I know! | 19:09 |
| paul_ | actually | 19:09 |
| paul_ | i'm not so sure it would | 19:09 |
| paul_ | :P | 19:09 |
| vb123 | It depends on the scenarios we want to light up. | 19:09 |
| nuclear_eclipse | paul_: it would first require building support for an AJAX-based UI model, and second would require a lot of coordination of some sort -- how do you handle two bugs created offline at the same time? who gets what bug id? | 19:10 |
| paul_ | hey, isn't gears a google project | 19:10 |
| nuclear_eclipse | yes | 19:10 |
| vb123 | Yes, the latter problem about ids we will have to support if we want to support rich clients with offline mode. There is the lease model where a client takes a range of bug ids to use. This however means non-ordered bug ids. | 19:12 |
| nuclear_eclipse | I'd rather just not support offline mode :P | 19:13 |
| paul_ | just generate a guid for a bug | 19:13 |
| paul_ | anyway, i'm not going to get distracted with sqlite | 19:14 |
| * paul_ hits nuclear_eclipse | 19:14 | |
| * nuclear_eclipse stabs paul_ in the kneecap | 19:15 | |
| vb123 | nuclear_eclipse: don't distract paul_ -- he is working on the rc :) | 19:16 |
| rioch | what do the different user access levels grant? | 19:16 |
| paul_ | vb123: john reminded me i'd coded a patch i'd forgotton about ::P | 19:17 |
| vb123 | so are we ready yet? | 19:17 |
| nuclear_eclipse | rioch: depends on how you configure all the access thresholds ;) | 19:17 |
| nuclear_eclipse | vb123: I've got two fix branches I'd like reviewed | 19:17 |
| vb123 | fixes for what? | 19:18 |
| rioch | nuclear_eclipse: :) its a fresh install, so whatever it is by default | 19:18 |
| paul_ | nuclear_eclipse: where? | 19:18 |
| nuclear_eclipse | bug 9744 and bug 10303 | 19:18 |
| mantisbot | Bug 9744 - seiji - open - assigned | 19:18 |
| mantisbot | Users behind proxy will get 'ERROR_SESSION_NOT_VALID' - http://www.mantisbt.org/bugs/view.php?id=9744 | 19:18 |
| mantisbot | Bug 10303 - jreese - open - assigned | 19:18 |
| mantisbot | Create and use string_api wrappers for multibyte string functions - http://www.mantisbt.org/bugs/view.php?id=10303 | 19:18 |
| nuclear_eclipse | both issues have a link to a public branch implementation | 19:19 |
| paul_ | url? | 19:19 |
| paul_ | in fact | 19:19 |
| paul_ | i've got issues with 10303 | 19:20 |
| paul_ | see what I pasted at you last night | 19:20 |
| nuclear_eclipse | paul_: you always have issues... | 19:20 |
| vb123 | You shouldn't have formatting in the language strings. | 19:20 |
| nuclear_eclipse | paul_: what you pasted at me last night had absolutely no context; I didn't know wtf I was looking at... | 19:20 |
| paul_ | replacing strlen->string_length fro every string is not appropriate imo | 19:21 |
| paul_ | i've been going through the ones that cause issues | 19:21 |
| paul_ | we already have utf8_strlen in git | 19:21 |
| paul_ | what I pasted you last night was a function to do a substr to nearest whole wordt | 19:21 |
| paul_ | that should support utf8 | 19:21 |
| nuclear_eclipse | paul_: I left alone stuff that I new wouldn't be needing utf8 | 19:22 |
| nuclear_eclipse | and utf8_strlen is not in git, and i don't think it should be | 19:22 |
| paul_ | I think we should support utf8 for people without mbstring | 19:22 |
| paul_ | if possible | 19:22 |
| nuclear_eclipse | paul_: see what I did for 10303 | 19:22 |
| nuclear_eclipse | I did do that | 19:23 |
| paul_ | no, you dont support utf8 unless mbstring is running | 19:23 |
| nuclear_eclipse | paul_: check 10303, *I DO* | 19:23 |
| paul_ | return strtoupper( $p_string ); isn't utf8 | 19:24 |
| vb123 | what does the secure session thing for? | 19:24 |
| nuclear_eclipse | vb123: prevents session hijacking by limiting the session to a single IP | 19:24 |
| paul_ | http://git.mantisforge.org/w/mantisbt/jreese.git?a=commitdiff;h=68debc492667bcfe6ad046c17171635439d9e76c | 19:24 |
| paul_ | and strtoupper is fine for all of those | 19:25 |
| nuclear_eclipse | without mb_*, I'm not sure how you could possibly support uppercasing UTF8 strings without reimplementing everything yourself.... | 19:25 |
| * paul_ has already done this exercise | 19:25 | |
| paul_ | in the case of the first strtoupper | 19:25 |
| paul_ | the captcha api only generates ascii strings atm | 19:25 |
| paul_ | in the case of the tags strtoupper | 19:25 |
| nuclear_eclipse | right, that was overlooked on my part | 19:25 |
| vb123 | nuclear_eclipse: where do you enforce the IP? | 19:25 |
| paul_ | the array that f_filter uses only supports ascii strings atm | 19:25 |
| nuclear_eclipse | vb123: in session_validate() | 19:26 |
| paul_ | in the case of the 3rd strtoupper, the array that f_filter only supports ascii strings | 19:26 |
| vb123 | nuclear_eclipse: this doesn't seem to be part of this patch. | 19:27 |
| nuclear_eclipse | ok, so the strtoupper stuff is unneeded atm, but the rest of it should be fine/needed | 19:27 |
| nuclear_eclipse | vb123: it was already implemented in Mantis since 1.1.3.... | 19:27 |
| nuclear_eclipse | that's what causes the proxy issue, because some proxies have multiple exit points | 19:27 |
| paul_ | nuclear_eclipse: seriously though, i've been going through using utf_* for utf functions | 19:27 |
| paul_ | you've got string to lowers on GET/POST | 19:28 |
| paul_ | my personal view is mbstring is going to be a dead module in 18 months or so (aka php6) | 19:28 |
| paul_ | it seems silly to add it now for stuff we dont need to add it against | 19:29 |
| nuclear_eclipse | paul_: I'm not sure why that's a problem... | 19:29 |
| nuclear_eclipse | and mb_strlen is quite useful in many places... | 19:29 |
| vb123 | Why is this a user option? | 19:29 |
| vb123 | I think it should be an admin issue | 19:30 |
| vb123 | Users shouldn't have to worry about proxy. | 19:30 |
| nuclear_eclipse | vb123: allows the user to disable it when they are behind a proxy, while allowing the rest of the users to maintain a more secure session | 19:30 |
| vb123 | Why don't we use a model like the cookie test? | 19:30 |
| nuclear_eclipse | vb123: because you can't rely on that | 19:30 |
| vb123 | is that possible? | 19:30 |
| vb123 | Do we need a separate cookie for that? | 19:31 |
| paul_ | nuclear_eclipse: i'd still rather go the approach we have been of looking at what causes a problem then trying to replace strlen->string_lower. | 19:32 |
| paul_ | As personally, I'd want to make sure that mantis works without mbstring | 19:32 |
| vb123 | I don't like the concept of an "unsecure" session. | 19:33 |
| paul_ | so for example the code at http://mantis.pastebin.com/d40fa8406 | 19:33 |
| paul_ | takes the fastest version of utf8_strlen i can find, faster utf8_substr | 19:33 |
| paul_ | and I believe implements a function to split a string by whole word | 19:33 |
| paul_ | which can be used for displaying relationships | 19:33 |
| nuclear_eclipse | vb123: the problem isfor example, user is behind load balancing proxy with two exit points A and B. the first 1, 5, 10, etc requests could come from node A, and then at random, the next request comes from node B; mantis can't tell if thats a legit request or a session hijack, so we leave it to the user to decide if they need to loosen security on their session | 19:35 |
| paul_ | nuclear_eclipse: what I was planning on doing with adding a comment to strlen's etc that dont require utf8 | 19:38 |
| nuclear_eclipse | my internet dropped at work -- only on phone atm... | 19:38 |
| paul_ | (hence why I can pretty much list ones which do/dont as I've spent a bunch of time looking ;) | 19:38 |
| paul_ | strlen/utf8_strlen seem an easy to remember way of doing it | 19:39 |
| paul_ | if the calls we're few enough i'd probably then be inclined to skip mbstring completely | 19:39 |
| nuclear_eclipse | iirc, mb_strlen is faster than strlen(utf8_decode()) | 19:39 |
| paul_ | correct by 0.002s to 0.008s or something | 19:40 |
| paul_ | but then, if we only need to call it once.. | 19:40 |
| nuclear_eclipse | so ithink it makes sense to prefer mb_* where available | 19:40 |
| nuclear_eclipse | and our code is littered with calls to strlen | 19:40 |
| paul_ | nod | 19:41 |
| paul_ | which is fine | 19:41 |
| paul_ | as the non-mbstring version of strlen is slower | 19:41 |
| paul_ | we need to change the ones which require utf8 to utf8_strlen | 19:41 |
| paul_ | we can't just change all | 19:41 |
| paul_ | and atm, I think there's a limited number we actually need to change | 19:41 |
| sshirley | I have another question concerning my migration from v1.1.0 to v1.1.8. On my_view_page.php on v1.1.0 under Resolved, each entry has an edit icon. But using the same DB on v1.1.8, none of those entries have that icon. Any idea? | 19:48 |
| nuclear_eclipse | check read only threshold in your configuration | 19:48 |
| paul_ | nuclear_eclipse: I think we need to look at the utf8 stuff properly/thoroughly | 19:49 |
| nuclear_eclipse | i cant imagine why it's that big of a deal to call mb* when the string isnt necessarily utf8... | 19:50 |
| paul_ | because strlen is easier to type | 19:51 |
| paul_ | the compatibility stuff is slower | 19:51 |
| paul_ | so we only want to use it on utf8 strings | 19:51 |
| nuclear_eclipse | it's not really that much slower... | 19:51 |
| paul_ | the point still remains, we've already dealt with a number of the functions that cause a problem | 19:52 |
| nuclear_eclipse | im not sure that youve even made a coherent point yet... | 19:52 |
| paul_ | strlen is a built in function to return length of string | 19:53 |
| nuclear_eclipse | right.... | 19:53 |
| paul_ | i've added utf8_strlen locally I think (if it's not already in core) | 19:53 |
| siebrand | [KK]Kirill: yes, I have a local mantis installation, just no e-mail config in there | 19:53 |
| paul_ | replace strlen with stringlower or something else is confusing | 19:53 |
| nuclear_eclipse | strlen -> string_length | 19:54 |
| siebrand | [KK]Kirill: basically we need to know if this is fixed in master, and the reason that it does not work on Mantisbt.org is that the tracker on there is on branch 'live' instead of 'master'. | 19:54 |
| paul_ | if strlen automatically does a slower utf8 version, i'd be incled to optimise it back to strlen for strings that we dont need to check utf8 for (for performance for people without mbstring) | 19:54 |
| nuclear_eclipse | paul_: so your point is you dont like the api names that i use... | 19:54 |
| siebrand | [KK]Kirill: if that assumption is true, it's a matter of time until it's fixed (I guess until 1.2.0 beta 1 is made available) | 19:54 |
| paul_ | no, my point is, we should do it where it's needed | 19:55 |
| paul_ | and review sensibly | 19:55 |
| paul_ | siebrand/myself stripped out a bunch of strtolower/upper's from displaying functions in favour of css | 19:55 |
| paul_ | email stuff phpmailer deals with utf8 | 19:55 |
| paul_ | for display of strings, our current string api functions deal with | 19:55 |
| nuclear_eclipse | so, you favor taking a whole bunch more effort to do less than what ive done using sed? | 19:56 |
| paul_ | yep | 19:56 |
| paul_ | (as i've been putting in that effort anyway) | 19:57 |
| nuclear_eclipse | api names i wouldnt mind changing, but i think youre insane otherwise =P | 19:57 |
| paul_ | from one point of view | 20:00 |
| paul_ | if i'm looking at adding non-mbstring utf8 safe functionality | 20:00 |
| paul_ | mbstring stuff can follow up as performance increase afterwards | 20:00 |
| paul_ | imo | 20:00 |
| paul_ | for example http://mantis.pastebin.com/d40fa8406 | 20:01 |
| paul_ | there's a substr there with a regex | 20:01 |
| paul_ | X thousand calls of utf8_substr2 takes 0.08s vs 0.002s for mbsubstring | 20:02 |
| paul_ | and we make about 30 substr calls a page or something | 20:05 |
| paul_ | if we change strlen->mb_strlen, we dont 'fix' out utf8 support, as we aren't fixing regex's/str_pad and other stuff that isn't utf8 | 20:06 |
| paul_ | there's no point changing database api to call string_lower as database table names need to be ascii (as adodb I believe calls strtolower internally) | 20:08 |
| nuclear_eclipse | gimme a bit to check the pastebins when my work access returns | 20:09 |
| paul_ | nod :P | 20:09 |
| paul_ | but you just happened to pick a topic to ask me about, that I just happened to have printed out and highlighted utf8 vs nonutf8 calls ;P | 20:09 |
| nuclear_eclipse | anywho, vb123, did you ever post an opinion of the session/proxy thing? | 20:11 |
| paul_ | my logic was basically, for the few places we need utf8, take the utf8 compat functions that phpbb/mediawiki use that have been tested by others and use those. We can then decide if we want to add mbstring for additional performance, or if we conclude the changes elsewhere in 1.2 speed things up and the number of calls we make are so low | 20:11 |
| paul_ | eof :) | 20:11 |
| paul_ | 20:38 < vb123> Why is this a user option? | 20:13 |
| paul_ | 20:38 < vb123> I think it should be an admin issue | 20:13 |
| paul_ | 20:39 < vb123> Users shouldn't have to worry about proxy. | 20:13 |
| paul_ | 20:39 < nuclear_eclipse> vb123: allows the user to disable it when they are behind a proxy, while allowing the rest of the users to maintain a more secure session | 20:13 |
| paul_ | 20:39 < vb123> Why don't we use a model like the cookie test? | 20:13 |
| paul_ | 20:39 < nuclear_eclipse> vb123: because you can't rely on that | 20:14 |
| paul_ | 20:39 < vb123> is that possible? | 20:14 |
| paul_ | 20:40 < vb123> Do we need a separate cookie for that? | 20:14 |
| paul_ | 20:42 < vb123> I don't like the concept of an "unsecure" session. | 20:14 |
| nuclear_eclipse | yeah, igot that, and responded, and never heard from him after that.... | 20:14 |
| nuclear_eclipse | hi again giallu | 20:24 |
| giallu | hi John | 20:24 |
| nuclear_eclipse | hmm, seems i may not have internet back onat work for some time now.... | 20:31 |
| nuclear_eclipse | oh fun, the Solaris and Windows machines have internet just fine, but not the Linux boxes... | 20:34 |
| nuclear_eclipse | hooray for chaining SSH logins.... | 20:35 |
| nuclear_eclipse | paul_: so remind me again why you're complaining about my wrapper of strlen() ? | 20:40 |
| nuclear_eclipse | and what's the purpose of utf8_substr2() ? | 20:41 |
| nuclear_eclipse | wwhat does that have anything to do with mantis and my patch to 10303 | 20:41 |
| paul_ | nuclear_eclipse: i've been adding/working on utf_* functions to support utf8 stuff | 20:48 |
| paul_ | and reviewing where we need utf8 functionality and where strlen and such suffice | 20:49 |
| nuclear_eclipse | I get that part, but can you answer the question? | 20:49 |
| paul_ | well, if i'm working out utf8_strlen and where to use it and where to just use strlen | 20:50 |
| nuclear_eclipse | in other words, why should we wait forever for you to review all these random things, rather than just take the patch I have and get 1.2cr out the door? | 20:50 |
| paul_ | the patch doesn't fix any 'bugs' though | 20:51 |
| paul_ | changing strlen isn't going to fix stuff if we haven't fixed the surrounding regex's | 20:51 |
| nuclear_eclipse | I don't see any harm in using a utf-capable strlen on "too many" things; if anything it just makes mantis run fractionally slower than it does now | 20:51 |
| paul_ | we've already reviewed strtoupper | 20:51 |
| nuclear_eclipse | ie, I don't know why we should even be trying to use strlen in one place and a utf8 strlen in other places? that just adds confusion | 20:52 |
| paul_ | strlen is for non-utf8 stuff for speed | 20:52 |
| nuclear_eclipse | facepalm | 20:52 |
| paul_ | utf8_strlen is for text | 20:52 |
| paul_ | strlen is a built in php function so is generally what people will use | 20:53 |
| nuclear_eclipse | it's really not that much of a speed difference between the two.... | 20:53 |
| paul_ | but the point still remains | 20:53 |
| paul_ | most of the existing issues in mantis relating to stuff | 20:53 |
| paul_ | and utf8 | 20:53 |
| paul_ | refer to our handling of regex's and strpad/substr | 20:54 |
| paul_ | (hence the utf8_substr function and strstop one) | 20:54 |
| nuclear_eclipse | my point is that we should just be using one consistent strlen call throughout our whole API so that you *don't* have to think about whether the data is utf or not, you just use a single function.... | 20:54 |
| paul_ | *longterm* that'll just be 'strlen' (aka php6+unicode support | 21:03 |
| paul_ | we need to review mantis properly for utf8 support (regardless of what we call string functions and how we do it) | 21:04 |
| paul_ | as mbstring doesn't magically fix regex's and other stuff | 21:04 |
| paul_ | so whilst your right in the sense that we could change the function name | 21:05 |
| paul_ | the fact we need to go through them anyway and see how they are used, strikes me as something easier to do 'as is now' | 21:06 |
| paul_ | (e.g. replace str* with utf8_str* if it requires utf8, add a comment if it doesn't require utf8 so we know we've considered it | 21:07 |
| nuclear_eclipse | I still don't see why we need to go through them all anyways; if we just change everything across the board to use utf8-compatible string functions, it'll work just as well, and both take less effort and be more consistent to work with | 21:07 |
| nuclear_eclipse | honestly, it seems like a maintenance and development nightmare to have to conditionally use different string functions based on the context of the content.... | 21:08 |
| paul_ | ok, obvious example | 21:08 |
| paul_ | we currently have: | 21:08 |
| paul_ | echo str_pad( '', strlen( $t_release_title_without_hyperlinks ), '=' ), '<br />'; | 21:08 |
| paul_ | with your code, we now have: | 21:09 |
| paul_ | cho str_pad( '', string_length( $t_release_title_without_hyperlinks ), '=' ), '<br />'; | 21:09 |
| nuclear_eclipse | right... | 21:09 |
| paul_ | now | 21:09 |
| paul_ | str_pad isn't utf8 safe | 21:09 |
| paul_ | iirc | 21:09 |
| paul_ | and iirc | 21:09 |
| paul_ | there isn't a mb_str_pad function | 21:09 |
| nuclear_eclipse | right, that's yet another evntual target to replace with mb* functions... | 21:09 |
| paul_ | there isn't a mb_str_pad function | 21:09 |
| nuclear_eclipse | so what? | 21:09 |
| nuclear_eclipse | how does that affect anything I've already done? | 21:10 |
| nuclear_eclipse | I never said my branch for 10303 was end-all-be-all of fixing utf8 issues | 21:10 |
| nuclear_eclipse | it's just the first step for strlen/strtolower/strtoupper... | 21:10 |
| nuclear_eclipse | but by rolling out these changes in a manner that's consistent across the entire codebase, it's easier to maintain and understand... | 21:12 |
| paul_ | I find utf_strlen / strlen easier to remember then string_length | 21:12 |
| nuclear_eclipse | eg, I shouldn't have to investigate the possible contents of a string just to figure which type of strlen to call; I should just have a single strlen type function to call | 21:12 |
| nuclear_eclipse | I don't give a damn what the function's called | 21:12 |
| paul_ | and b using 2, and changing them as we investigate whether the use requires utf8 | 21:12 |
| paul_ | we know what we've covered | 21:13 |
| nuclear_eclipse | sed know's that I've covered everything! | 21:13 |
| * paul_ sighs | 21:13 | |
| nuclear_eclipse | the only thing I haven't covered is admin/* | 21:13 |
| nuclear_eclipse | and that's because admin/* doesn't use the standard api setup | 21:13 |
| paul_ | I meant, what we've covered in terms of functionality | 21:13 |
| nuclear_eclipse | paul_: you're missing my whole point... | 21:14 |
| paul_ | I know your point | 21:14 |
| paul_ | I think it's silly to call random function that might return the length of the string, when there's a strlen function which does it | 21:14 |
| nuclear_eclipse | why the hell do we need to investigate each usage of a function when we can just replace all calls to strlen() with wtfyw_strlen()? | 21:14 |
| paul_ | the silly thing is that php's strlen doesn't return the correct length of a string due to php not supporting utf8 ;/ | 21:14 |
| nuclear_eclipse | paul_: when mantis requires php6, then we just sed utf8_strlen back to regular strlen; until then, by consistently using utf8_strlen everywhere, we don't have to figure out what a string uses first, we just use a single function guaranteed to work... | 21:16 |
| * nuclear_eclipse heads home | 21:17 | |
| paul_ | argh | 22:16 |
| paul_ | nuclear_eclipse: on the offchance, lurking? | 22:41 |
| nuclear_eclipse | perhaps... | 22:41 |
| paul_ | relationship api | 22:42 |
| paul_ | with bugobject stuff | 22:42 |
| paul_ | should api take in a bug id or the object? | 22:42 |
| nuclear_eclipse | not familiar with the relationship api, but assuming it doesn't actually need data other than the bug id, just pass an id | 22:44 |
| paul_ | wtf | 22:59 |
| paul_ | nuclear_eclipse: have you ever looked at bugnote api? | 23:00 |
| nuclear_eclipse | only slighty | 23:00 |
| nuclear_eclipse | looked at a subsection of it recently to fix bugnotes in email notifications | 23:00 |
| paul_ | I hate looking at stuff :( | 23:01 |
| paul_ | generates work | 23:01 |
| nuclear_eclipse | it wouldn't generate work if you could just look at it and ignore the urge to rewrite things... | 23:02 |
Generated by irclog2html.py