| ../irclogs/#mantishelp.2009-03-28.log | ||
| --- scribe started --- | 00:00 | |
| nuclear_eclipse | siebrand_alt: you there? | 01:38 |
|---|---|---|
| dhx_m | hi | 03:28 |
| nuclear_eclipse | howdy | 03:28 |
| dhx_m | I was just wondering if someone could help me understand git usage a little better | 03:28 |
| dhx_m | I've setup a mantis fork @ http://git.mantisforge.org/w/mantisbt/dhx.git | 03:28 |
| nuclear_eclipse | I can try :) | 03:29 |
| dhx_m | which is set to push mode | 03:29 |
| dhx_m | I guess I need to clone (?) that empty repository to my local computer | 03:29 |
| dhx_m | as per http://docs.mantisbt.org/master/en/developers/dev.contrib.clone.html | 03:29 |
| dhx_m | and then checkout the mantis-git master to my local branch | 03:30 |
| dhx_m | and commit those changes to my own branch? | 03:30 |
| nuclear_eclipse | no, you need to clone the official repository, and then push to your private fork | 03:30 |
| dhx_m | so my "working copy" is the official repository? | 03:31 |
| dhx_m | (I come from an SVN background... but desperately want to get up to speed on git because the theory behind it is much superior) | 03:31 |
| nuclear_eclipse | yes, if you have a clone of the official repo, you can run `git remote add <name> <push_url>` to add a remote for the other repository | 03:32 |
| nuclear_eclipse | http://www.eecs.harvard.edu/~cduan/technical/git/ | 03:33 |
| nuclear_eclipse | that will likely help explain a lot of Git for you | 03:33 |
| dhx_m | thanks I'll give it a read | 03:33 |
| dhx_m | I guess that I create a branch for each feature change I made to mantis | 03:33 |
| dhx_m | but does that mean I need to have 1 git repository for each branch? | 03:34 |
| dhx_m | ie. I'd need to fork mantis-git for each 'branch'? | 03:34 |
| nuclear_eclipse | for a simple set of patches, it might be simpler to skip setting up a private fork, and just use formatted patches, but a fork repo can house multiple feature branches, so if you're working on a lot of separate contributions, a private fork on mantisforge is more useful/versatile | 03:34 |
| dhx_m | yep that is what I'll be doing | 03:35 |
| dhx_m | I have a few patches that I'm trying to contribute to Mantis | 03:35 |
| nuclear_eclipse | with git, it's better to have all of your branches in the same repository, rather than making a separate clone for each feature | 03:35 |
| dhx_m | and they're getting hard to separate within one directory | 03:35 |
| dhx_m | ok | 03:35 |
| dhx_m | so the structure is something like http://git.mantisforge.org/w/mantisbt/dhx.git/branch1/ | 03:36 |
| dhx_m | http://git.mantisforge.org/w/mantisbt/dhx.git/branch2/ | 03:36 |
| dhx_m | etc? | 03:36 |
| dhx_m | that is what I'm used to with SVN | 03:36 |
| nuclear_eclipse | not quite | 03:38 |
| dhx_m | actually that tutorial you sent me explains it | 03:39 |
| nuclear_eclipse | git tracks branches outside of the "working directory", so when you set up other branches, it's only known about in the .git until you specifically switch to that branch, at which point it replaces your current workspace | 03:39 |
| dhx_m | aha :) | 03:40 |
| dhx_m | that makes sense now | 03:40 |
| dhx_m | thanks | 03:40 |
| nuclear_eclipse | I'd suggest using a separate branch for each feature you work on | 03:40 |
| dhx_m | so they're sort of 'invisible' and you switch between them within the same working directory | 03:40 |
| nuclear_eclipse | correct | 03:40 |
| dhx_m | rather than the SVN approach of having 1 working directory per branch | 03:40 |
| nuclear_eclipse | same thing with tags, they're just a pointer to a commit, and don't really show up unless you specifically check them out in place of a branch | 03:41 |
| dhx_m | so if you switch to a branch, make some changes... and don't commit | 03:42 |
| dhx_m | then change to another branch, you lose changes | 03:43 |
| dhx_m | ? | 03:43 |
| nuclear_eclipse | actually, git will generally prevent you from doing destructive things like that unless you explicitly tell it to override the safeguards | 03:43 |
| nuclear_eclipse | so if you have uncommitted changes, git won't let you switch branches | 03:44 |
| nuclear_eclipse | unless of course you add the -f/--force option to ignore uncommitted changes | 03:44 |
| dhx_m | ah nice :) | 03:45 |
| dhx_m | my only other question is that when submitting patches, do I just tell someone to pull one of my branches? | 03:45 |
| dhx_m | as opposed to attaching a manual patch? | 03:46 |
| nuclear_eclipse | git can generate a 'pull request' that sums up the changes in your branch, or it can generate 'formatted patches' that contain author and commit info and con be easily applied to a repository | 03:47 |
| nuclear_eclipse | both methods are covered in the developer's guide | 03:47 |
| dhx_m | I understand now :) | 03:47 |
| dhx_m | thanks | 03:47 |
| dhx_m | I'll go out and give it a try | 03:47 |
| nuclear_eclipse | you're welcome, and good luck :) | 03:47 |
| vb12345 | nuclear_eclipse: the IRC link appears for anonymous users. Clicking it gives access denied. It should be hidden for users who don't have access. | 04:10 |
| dhx_m | just a question about how I set this up... | 04:23 |
| dhx_m | I take it I clone mantis-git | 04:24 |
| dhx_m | and push/merge that repository into dhx-git | 04:24 |
| dhx_m | then I create a new local directory from which I clone dhx-git to | 04:26 |
| dhx_m | and I do all my changes in dhx-git | 04:26 |
| dhx_m | and pull updates from mantis-git when needed? | 04:26 |
| dhx_m | most of the tutorials I read assume you work on a local repository | 04:27 |
| dhx_m | not a remote one | 04:27 |
| --- scribe started --- | 17:56 | |
| siebrand | giallu: thanks. That was helpful. Now I was able to close a few issues :) | 18:49 |
| nuclear_eclipse | siebrand: just as an fyi; use "fixes #xxx" in your commit message, so that the Git integration for Mantis will auto-link issue IDs to the changeset | 18:50 |
| siebrand | right. Thanks. | 18:51 |
| bartik | Hello. May I ask some mantis code related questions ? | 18:53 |
| siebrand | bartik: yes | 18:54 |
| bartik | I have implemented local avatars for the 1.1.6 and I have stumbled across this two functions file_delete_attachments and file_delete_project_files | 18:55 |
| bartik | Is there a particular reason these are not coded as one function ? | 18:56 |
| bartik | I mean the difference is the bug_id and project_id in the select query. | 18:56 |
| bartik | Well this somewhat touches on the main reason me is here. I wanted to ask if there is interest in having the local avat feature. | 18:57 |
| bartik | !ping | 18:59 |
| giallu | bartik, yes. ther is a bug open for that | 18:59 |
| bartik | bug ? | 19:00 |
| bartik | Ah I understand you treat everything as bug even a feature I suppose. | 19:00 |
| giallu | yeah | 19:01 |
| bartik | Well I have an implementation but it is not the head revision. But as I browse the repository it doesnīt seem very different from what I have implemented | 19:01 |
| giallu | let's see | 19:01 |
| bartik | So are there any particular requirements ? | 19:02 |
| bartik | I mean coding preferences whatever ? I have tried to reuse as much code as possible. | 19:02 |
| giallu | bug 8257 | 19:03 |
| mantisbot | Bug 8257 - mkornatzki - open - acknowledged | 19:03 |
| mantisbot | store avatar local - http://www.mantisbt.org/bugs/view.php?id=8257 | 19:03 |
| giallu | have a look there and tell me what you think about: | 19:03 |
| giallu | 1. the patches I see there | 19:03 |
| giallu | 2. how that compare with yours | 19:04 |
| giallu | then, about your questyion before about api | 19:04 |
| giallu | if you find stuff in need of a clean up, feel free to propose your improvements | 19:05 |
| giallu | an the file api actually would need some love... | 19:06 |
| bartik | I have now looked over the change. I have a slightly different approach. | 19:09 |
| bartik | For the avatar. For me the avatar is handled similarly to the attachments. | 19:09 |
| bartik | I have added a new table called mantis_avatar_file_table. Instead of a bug_id or project_id I use the avatar_id | 19:10 |
| bartik | This way I was able to reuse the download_file.php for loading avatars and those file_delete_project_files etc. (hence the question regarding these) | 19:11 |
| bartik | But there is no resizing in my version. On the other hand you can store the avatar either in database, file or ftp as any other attachments. | 19:12 |
| bartik | The selection of what avatar to use (gravatar or local) is of course configurable but alas not per user. | 19:14 |
| giallu | bartik, attach your patch on 8257, I'll have a look | 19:15 |
| giallu | but please note we usually do not introduce new features on the stable serie | 19:16 |
| giallu | so it needs to apply on git master | 19:16 |
| bartik | I understand. Thatīs why I was asking. I didnīt know about the existing version. I could maybe save some effort :-) | 19:18 |
| giallu | ok | 19:19 |
| giallu | so your implementation use a global setting to decide if avatars are local or gravatar? | 19:21 |
| bartik | yes. I have never thought about using this in "mixed" mode. You would need to introduce a preference setting in user profile (which I didn't) to have this but shouldn't be hard. | 19:23 |
| bartik | Another question regarding the passwords. Was there any thought about checking some minimal password strength ? | 19:26 |
| bartik | In the account_update.php like adding a "_default_" function which could be overwritten by a "_override_" function ? | 19:27 |
| bartik | My "hack" is pretty ugly with that if (... sitting in the middle of the code. | 19:28 |
| giallu | another open bug somewhere... | 19:31 |
| giallu | but now, dinner time :) | 19:31 |
| giallu | I'll be back later | 19:31 |
| bartik | me be here | 19:32 |
| bartik | well the pages about the code checkout seem to refer to svn but your repository is git now am I right ? | 19:52 |
| giallu | back | 21:13 |
| giallu | bartik, yes, we use git now | 21:13 |
| bartik | quite a time consuming dinner... | 21:13 |
| giallu | eheh | 21:13 |
| giallu | add 2 children in the mix | 21:14 |
| bartik | mine 2 are already sleeping :-) | 21:14 |
| giallu | here too, but I didn't rush to pc :) | 21:14 |
| giallu | so | 21:14 |
| giallu | did you find how to get git code? | 21:15 |
| bartik | I have already cloned the main git but now trying to find out how to get the master for 1.1. | 21:17 |
| bartik | using tortoisegit | 21:17 |
| bartik | ok from the command line it works. freakin gui... | 21:29 |
| bartik | after a short fight with setup itīs set up :-) | 21:48 |
| bartik | on the other hand there is a feature iīm going to explore but maybe I ask first. Do you have gnupg encryption for email ? | 21:52 |
| siebrand | Does mantis allow multiple users with the same e-mail address? | 22:39 |
| siebrand | yes, it does. | 22:40 |
| siebrand | (testing proved to be quicker :P) | 22:40 |
| siebrand | giallu: how about we make mb_string a requirement for 1.2? | 23:29 |
| siebrand | it's hard to count Cyrillic characters without it... | 23:30 |
| siebrand | I think we just need it for proper i18n support. | 23:30 |
| nuclear_eclipse | siebrand: I'd prefer to just extend what I've done for string_lower() where we use mb_* if it's available, and fall back to regular functions otherwise; no need to penalize those of us who don't need unicode and don't/can't use mb_* | 23:40 |
| siebrand | nuclear_eclipse: I found another solution for the problem I was working on. utf8_decode() appears to do the trick. | 23:41 |
| siebrand | Hmm, running into "The filter you are trying to use is too old to be upgraded. Please re-create it." | 23:43 |
| siebrand | That's interesting, as I just created it. | 23:43 |
Generated by irclog2html.py