Differences

This shows you the differences between two versions of the page.

--- development:communication:archives:100302_weekly [2010/03/01 18:45]
j_k9 improved legibility
+++ development:communication:archives:100302_weekly [2010/04/19 19:29] (current)
j_k9 fixed syntax
@@ Line 7: / Line 7: @@
 This meeting's agenda is available [[development:communication:agendas:100302_weekly|here]].
-----
+===== Summary =====
+As sent to the Mira Development Mailing List on 02/03/2010 with the subject: “Summary of 02/03/2010 Meeting”
+> Hi all,
+>
+> Tonight's meeting was attended by Alan, Neeraj and myself (and Thurston, through the wires). You can find a transcript of the meeting here:
+> http://miragroupware.org/wiki/doku.php/development/communication/archives/100302_weekly
+>
+> This meeting was particularly important because we are approaching our very first release of Mira Groupware and we are trying to tie together all the loose ends before we freeze the codebase and concentrate on testing it.
+>
+> We first discussed the merging of the separate code development branches into 'trunk'.
+> Alan's mira-utilities branch is almost ready to be merged, but he would like to implement a GUI form to create new Workplaces before merging the branch. That is the only outstanding major task he has left to implement in his branch for Mira 0.1.
+> Neeraj's trunk_client_parsing branch contains working code for Client-side encryption of a user's password before it is transmitted. We discussed the merits of using a unique, randomly generated salt and decided that we would use this Client-side encryption (with a hard-coded salt) in Mira 0.2 as an extra layer of security during the authentication process. In the future, the entire network communication between Mira Client and Mira Server will be secured through an implementation of Transport Layer Security (TLS).
+>
+> The next topic was the cleaning up of old code branches which are assigned to our project on Launchpad. The only branch which is no longer active is Thurston's client-gui branch because its code has been merged into Alan's mira-utilities branch. As a result, I have labelled this branch "Merged" so that further development on it should not be expected.
+>
+> Our next task was to identify the Current Release Holds. As they stand, these are:
+>   * GUI - Workplace creation form (Alan is working on)
+>   * GUI - Update user totals in Chat widget groups (Alan is working on)
+>   * Network - Segmentation fault bug during login process (Neeraj is working on)
+>   * Boost - Property tree error with MingW (Alan to look into)
+> I have updated the Roadmap on the wiki to reflect these.
+>
+> We then discussed the network messages to be used for the three different communication systems: the Chat widget, the Workplace-wide Instant Messaging Utility, and the email-like Message Center.
+> The Instant Messaging Utility will use its own Utility message.
+> The Chat widget will use the M message (this message and its parsers will be modified in Mira 0.2 to suit the requirements of the Chat widget).
+> The Message Center will use the PM message (this message and its parsers will be modified in Mira 0.2 to suit the requirements of the Message Center).
+>
+> Another feature to be added in Mira 0.2 is tab support in the Chat widget. At the moment, the widget's text browser updates itself to show the relevant conversation whenever you select a Member from one of the chat groups. This will be changed to a separate tab for each conversation (to each Member) to make the communication experience more intuitive and to allow for features such as flashing tabs upon the receipt of a new message.
+>
+> We then reviewed the status of Thurston's mira/installer branch, which contains the installer files he built using BitRock's InstallBuilder (and his own expertise). Apparently, Thurston's code won't need to be amended in order to create installers for the latest versions of Mira Client and Mira Server - it will work as is. I intend to test this and report to the mailing list with the outcome.
+>
+> Please keep up the great work!
+>
+> Regards,
+>
+> Max Bossino
+> Project Manager
+> http://miragroupware.org
+===== Transcript =====
 [00:47] j_k9: Hello clsk, neeraj :)
 [00:47] neeraj: Hi Max and Alan. How are you both?
 [00:48] j_k9: Good thanks, and yourself?
 [00:50] neeraj: good. I am going to through the email exchange between you guys last week.
 [00:56] clsk: hello
 [00:59] j_k9: Hi clsk :)
 [00:59] j_k9: Let's wait another five minutes in case anyone else turns up
 [01:00] clsk: Max, I'm sorry I didn't respond to your email
 [01:00] clsk: i just read it
 [01:00] clsk: been sick since last night
 [01:00] j_k9: Don't worry about it - I hope you're feeling better! Are you ok to be on here now?
 [01:01] clsk: yea, it's fine
 [01:03] neeraj: Does anyone know where to set -g option for the getting debug information? So that I could use gdb for server code.
 [01:04] neeraj: ok got it .. thanks
 [01:04] j_k9: Sorry Neeraj, not my area of expertise - I just used 'gdb ./mira_server' and then 'run' :)
 [01:05] j_k9: Has everyone read the agenda?
 [01:05] j_k9: http://miragroupware.org/wiki/doku.php/development/communication/agendas/100302_weekly
 [01:05] neeraj: yeah I saw in cmake file the option is already set to -g
 [01:05] clsk: yep -g should already be set
 [01:05] neeraj: so I can use gdb right away
 [01:06] clsk: on both the client and server
 [01:06] neeraj: ok I didn't know
 [01:06] clsk: it's in the compilerOptions.cmake file
 [01:06] neeraj: Yeah I have gone over the agenda
 [01:06] clsk: yep, I've read the agenda
 [01:06] neeraj: Thanks I got it.
 [01:07] j_k9: Great. Let's start the meeting, because it looks like no-one else is going to join.
 [01:08] j_k9: Regarding the merging of branches into trunk, Alan: is the code in mira-utilities ready to be merged? Or are we waiting to fix the Seg fault before merging it?
 [01:08] clsk: we could do it before or after.
 [01:08] clsk: but I'm still working on something
 [01:09] clsk: I'm going to create a form to create new groups
 [01:09] j_k9: Ah, ok. That sounds like a good idea - it will help for testing purposes (without having to do it the long way, through telnet).
 [01:10] clsk: right
 [01:10] j_k9: I think you should merge your branch once you've finished working on that task. Are there any other pending tasks for your branch?
 [01:10] clsk: hm not really
 [01:10] j_k9: In that case, by all means go ahead. :)
 [01:10] clsk: I'd have to look at my notes. I'm running windows ATM
 [01:11] clsk: all my notes are on my linux installation
 [01:11] clsk: but there isn't anything major pending
 [01:11] j_k9: Would you please send me a copy of them, when you get the chance? Or if not to me, to the list, so that everyone can have an idea of the timeframe left for your branch?
 [01:12] clsk: sure
 [01:12] j_k9: Neeraj: Did you make any progress in your trunk_client_parsing branch?
 [01:12] j_k9: https://code.launchpad.net/~neeraj-rajgure/mira/trunk_client_parsing
 [01:13] neeraj: If there is nothing else we could think of then I have to get the three way communication done foe creating new user
 [01:14] neeraj: I was not in the favor of sending the salt  from server to client in the plain text
 [01:14] j_k9: Hmm... That's probably a good temporary solution. Perhaps in the future we could use SSL tunnelling to secure the connection?
 [01:15] neeraj: I think instead of sending the salt from Server to Client why don't we hard code  it on both the ends?
 [01:16] j_k9: But wouldn't hard-coding the salt defeat the purpose of having a randomly generated, unique salt?
 [01:16] neeraj: maybe not because we are open source anyone who has source could look at it at decrypt other peoples information.
 [01:17] neeraj: security wise sending clear text salt from server to client isn't good solution either.
 [01:17] j_k9: http://codeeleven.blogspot.com/2007/11/there-is-no-point-in-client-side.html
 [01:18] j_k9: Admittedly, it's talking about a web application, but we could apply the same to our platform.
 [01:20] j_k9: To be honest, I think there *is* some value in encrypting on the client-side, in any case. But I think our best bet is to look into TLS in the future for greater security
 [01:20] clsk: the idea is to use SSL in the future
 [01:21] clsk: that's what we've always talked about
 [01:21] clsk: I just haven't had the time to implement it
 [01:21] neeraj: ok then which way to go?
 [01:22] neeraj: Thanks for the post Max.
 [01:22] j_k9: I think it's worth encrypting it on the client-side in any case because if, for example, the password were read in cleartext during its transmission to the server, that password could then be used to hijack the user's accounts on other systems (most people use the same password on multiple systems).
 [01:23] neeraj: make sense. Thats where I stopped :(
 [01:23] j_k9: So the way forward, in my opinion, is to enable client-side encryption of the password (with either a unique salt or a hard-coded salt - that's up to you), and then to use TLS.
 [01:24] j_k9: Why don't you use a hard-coded salt? It would be easier to implement. If we then see a significant benefit to using unique salts, in the future, we could decide to implement that instead. How does that sound, Neeraj?
 [01:25] neeraj: Currently it is kind of hard-coded
 [01:25] j_k9: Have you tested it? Does it work?
 [01:25] neeraj: but as we are open source anyone can read the code and know what we are doing so I got concerned about it.
 [01:26] neeraj: yeah I had tested it before I post the link to you guys
 [01:27] j_k9: Great, in that case that's another feature that's complete. clsk: Do you think we should merge this into trunk for 0.1 or should we wait until after the 0.1 release? I wouldn't want there to be (difficult-to-resolve) conflicts with the authentication code and the login code in mira-utilities
-[01:28] clsk: neeraj: This is only temporary. By the time we release a stable version we'll be using SSL
+[01:28] clsk: neeraj: This is only temporary. By the time we release a stable version we'll be using SSL.
 [01:28] clsk: j_k9: I'd like to leave it for 0.2
 [01:28] j_k9: clsk: That make sense to me.
 [01:29] neeraj: clsk, so should I leave the way it is hard coded?
 [01:30] clsk: neeraj: yes
 [01:30] j_k9: Sorry, before we move on, can I just ask clsk: have you merged Neeraj's Client-side message parsers into your code? (If I'm not mistaken, they're needed for the Workplace creation routine.) I'm talking about the code he added in Revision 192 ("Added parsing code for GROUP_ACCEPTED, GROUP_DESCRIPTION and USER message for server response.")
 [01:33] clsk: j_k9: I haven't. That code doesn't really do anything ATM
 [01:33] j_k9: clsk: Is it code that you will need for 0.1?
 [01:34] clsk: no
 [01:34] j_k9: Ok, thank you.
 [01:35] j_k9: Let's move on to our next topic (which I've sneaked into the agenda): cleaning up the branches assigned to our Launchpad project. There only seems to be one which is defunct, which is Thurston's client-gui branch. https://code.launchpad.net/~mira-dev/mira/client-gui
 [01:35] j_k9: Is this still in use? Does it have untested code which needs to be merged into trunk or is it all legacy code?
 [01:38] clsk: all the code in client-gui is currently on my branch as well
 [01:39] j_k9: Ok, so we can change that branch's status to 'Merged' to denote that it doesn't expect any further activity.
 [01:41] j_k9: The next point on the agenda is the tidying up that needs to be done before pushing the 0.1 release, which is a longer way of asking what our release holds are before we can freeze and test the code.
 [01:41] j_k9: Alan, you've said that you want to implement a Workplace creation form. We also need to fix the Seg fault bug. Is there anything else outstanding?
 [01:42] clsk: nop
 [01:42] clsk: those are the only two things
 [01:42] neeraj: i am looking at the login bug is that the same as seg fault?
 [01:42] j_k9: Neeraj: Yes, that is the same bug.
 [01:43] j_k9: Have you had any success with it?
 [01:43] neeraj: No I rather just started looking at it before the meeting.
 [01:43] neeraj: The code wasn't compiling till yesterday :(
 [01:44] clsk: oh forgot. There is one more outstanding issue
 [01:44] j_k9: That's ok. :) It would be great if you could take a look into it and try to get to the heart of the problem. If you do, please try to fix it. clsk: Should Neeraj push his fix to your branch or to a separate branch?
 [01:44] clsk: the property tree with mingw
 [01:45] j_k9: Ah, yes. Shilpa said she had trouble compiling your branch because of that.
 [01:45] clsk: I'm going to have to install mingw here to see if I can get it working
 [01:45] clsk: that's why I booted into windows today
 [01:46] clsk: I'll see if I can get a dev environment going with mingw tonight
 [01:46] clsk: depends on how I feel later on tonight
 [01:46] j_k9: Is this a bug in Boost or in our usage of the Property tree?
 [01:46] clsk: but it might take me a while because I have a dev environment already set up with visual c++
 [01:46] clsk: j_k9: in boost
 [01:47] clsk: if the code compiles on linux it should compile everywhere else
 [01:47] neeraj: Me too was facing that problem there so reinstalled ubuntu
 [01:47] j_k9: clsk: Is there any chance the Boost community could help? They seem to be pretty responsive: http://groups.google.com/group/boost-list/topics
 [01:47] clsk: yes
 [01:47] clsk: they are actually.
 [01:47] clsk: but I can't ask for help without being able to provide more input on the issue
 [01:48] j_k9: clsk: Would you please take responsibility for looking into this? You are without a doubt the one most familiar with that library (and your code).
 [01:48] j_k9: Obviously, once you're feeling better. :)
 [01:49] clsk: yep
 [01:49] clsk: that's what I was planning on doing
 [01:50] j_k9: Great. Ok, now that we've talked about this, I'll update the Release Holds on the wiki after the meeting.
 [01:51] j_k9: The next point is a design issue which we don't *have* to discuss now (it can wait until 0.2), but we can if you have the time. It's about the M and PM messages and which we should use for the three communication systems: the Chat widget, the Workplace-wide Instant Messaging Utility, and the email-like Message Center.
 [01:51] j_k9: Does anyone object to discussing this now?
 [01:51] clsk: well the workplace-wide Utility will be a Utility message
 [01:52] j_k9: clsk: Right, so that would be nothing to do with the M and PM messages.
 [01:52] clsk: right
 [01:52] j_k9: So that leaves the other two systems. Should we consider using M for the Chat widget and PM for the Message Center, in 0.2?
 [01:52] clsk: yep
 [01:52] clsk: that'd make sense
 [01:53] j_k9: Ok. We'd have to redesign the protocols to suit the two systems and then reimplement the parsers, but I don't think that'd be too much work.
 [01:55] clsk: right
 [01:55] j_k9: The next discussion topic is the recent developments in the code branches. Alan, I think we're looking at you for this :) Do you have any work left to be done on the Chat widget?
 [01:55] clsk: nop
 [01:55] clsk: I corrected the issue you talked about
 [01:55] clsk: about sending a message when no user is selected
 [01:56] j_k9: clsk: That's great, thanks.
 [01:56] j_k9: Is there any chance we could also update the numbers in each group? That seems to be hard-coded at the moment.
 [01:56] j_k9: e.g. Offline (2/5)
 [01:56] clsk: oh yea
 [01:56] clsk: I had forgotten about that
 [01:56] clsk: that shouldn't be too much of a problem
 [01:56] j_k9: Would that take too long?
 [01:56] clsk: nop
 [01:57] j_k9: Excellent.
 [01:58] clsk: hm
 [01:58] clsk: I need to write that down somewhere
 [01:58] j_k9: As for the tabs idea, that's something we could look into for 0.2. Perhaps Thurston's code for the Utility tabs could help with that, when we get there?
 [01:58] j_k9: clsk: I'll send round a summary after this meeting. :)
 [01:58] clsk: cool thanks
 [01:58] clsk: yea, that's something for 0.2
 [01:59] clsk: hm I don't think it should be hard to implement because of the way the chat widget is currently implemented
 [01:59] clsk: I use a different text browser for every user. So just changing that to use a tab for every user should be easy
 [02:00] j_k9: clsk: That's great. That will make it slightly more intuitive to navigate the conversations, and we could have a tab flash upon the receipt of a new message, for example.
 [02:00] clsk: yep. I'm not sure how to do that yet though
 [02:02] j_k9: Perhaps Thurston can help with it, if he has some more free time, when we start working on 0.2.
 [02:03] j_k9: Our next discussion topic is the question of building installers for Mira Client and Mira Server once all the code for 0.1 has been merged into trunk and thoroughly tested.
 [02:04] j_k9: I know Thurston was working on installers (for both Windows and Linux for Mira Client) in a separate branch, but I'm not sure to what extent it will need to be amended to suit our updated codebase.
 [02:04] j_k9: His branch is: https://code.launchpad.net/~tstone2077/mira/installer
 [02:07] clsk: hm I'm not really sure either
 [02:07] clsk: I'm not very familiar with the installer
 [02:07] clsk: I'd have to read the documentation
 [02:07] clsk: I plan on doing that after we've tested our code
 [02:09] j_k9: clsk: Ok. When we get to that stage, feel free to get in touch with me for the Open Source serial licensed to Mira.
 [02:09] j_k9: I know I sent it to Thurston, but I'm not sure I sent it to you.
 [02:10] clsk: I don't think you did
 [02:10] clsk: if you did it should be archived on my email
 [02:11] j_k9: Actually, I'll just forward you the email now, just in case you don't have it.
 [02:12] j_k9: Done. :)
 [02:13] j_k9: The final discussion topic was about the Seg fault (the login bug), but Neeraj: you're going to work on this, correct?
 [02:13] neeraj: yeah I was looking at it now
 [02:14] neeraj: and sees so is_logged_in() returns false
 [02:14] neeraj: hopefully I get it sorted out soon .
 [02:14] j_k9: Ok, that's great! Does anyone have anything else they would like to talk about, or any issues to raise?
 [02:14] clsk: I'm talking to Thurston on msn messenger at the moment
 [02:15] clsk: to find out how complete his work on the installer is
 [02:15] j_k9: Ok
 [02:15] j_k9: Please let him know that we say hi :P
 [02:18] clsk: I was working on a branch with it... i had a bit of restructuring goin on
 [02:18] clsk:  so there is a mira-installer directory along side of the client and server directory
 [02:18] clsk: the installer is ok... it could use some more work, but it will install
 [02:19] clsk: it will install the executables, but it won't install the service
 [02:19] clsk: this is what he says
 [02:19] clsk: I think that should be enough for now
 [02:20] j_k9: clsk: By "it will install the executables, but it won't install the service", do you mean that it will install Mira Server as an executable as opposed to a service? (I presume we're not considering Mira Client as a service)
 [02:20] clsk: right
 [02:20] clsk: that's what I assume too
 [02:21] j_k9: Ah, ok. And do we need to modify the InstallBuilder configuration files or should that just work off the CMake configuration/makefiles?
 [02:21] clsk: from what he says it should just work
 [02:21] j_k9: That's great. :)
 [02:21] clsk: We'd have to check
 [02:22] j_k9: Yep. If you'd like, I can give it a go on Ubuntu to see it if works.
 [02:22] clsk: sure
 [02:22] j_k9: That saves you from having to test it yourself :)
 [02:22] j_k9: I'll send an email to the list with the outcome.
 [02:23] j_k9: Ok, so is there anything else we'd like to discuss?
 [02:25] j_k9: Actually, there is one thing I'd like to add - I've just encountered a bug with the Chat widget, clsk. I have added another user called "test" who is offline (along with the user "clsk"). I am logged in as the user "max". I send a message to clsk, then click on test and send a message to test. I then click on clsk to send a message, but the Chat window doesn't change.
 [02:25] j_k9: Have you come across this bug before?
 [02:26] clsk: hm weird.
 [02:27] clsk: If it's what I think it's something I've fixed, but haven't pushed yet.
 [02:28] j_k9: If you'd like, I can email you some screenshots of what's happening. In the terminal (which shows in the background), it looks like the messages are going through just fine but the GUI isn't updating.
 [02:28] clsk: Could you send me an email with the steps you did so I can try to reproduce it?
 [02:28] j_k9: Sure thing.
 [02:29] clsk: thanks
 [02:29] j_k9: Ok, is there anything else to discuss? :)
 [02:30] clsk: not that I can think of
-[02:30] neeraj: nothing at my end ...
+[02:30] neeraj: nothing at my end ..
 [02:31] neeraj: Just to update ... there is a issue with TcpConnection
 [02:31] neeraj: TcpConnection object I man
 [02:31] neeraj: mean *
 [02:31] j_k9: Neeraj: Have you found the source of the error?
 [02:32] neeraj: not really but I think I know what is happening
 [02:32] neeraj: I will keep you guys posted. Tomorrow and day after is a tight day for me though
 [02:33] j_k9: Whenever you can find the time would be great :) Actually, if you could post a reply to the Login bug thread on the mailing list whenever you have an update - such as now - that would be great! That way, everyone will be aware of your progress on this bug.
 [02:34] neeraj: sure
 [02:35] j_k9: Ok, I think we can call this meeting to an end. Many thanks to both of you for joining! I'll send a summary to the mailing list as soon as I can.
 [02:36] neeraj: Thanks to both of you too.
 [02:36] neeraj: gn
 [02:36] j_k9: Goodnight Neeraj