User:Jeko: Difference between revisions

From SpaceElevatorWiki.com
Jump to navigationJump to search
(Undo revision 2964 by RuthLindquist (Talk))
Line 1,016: Line 1,016:


1h. Page footer (showing blog "links", search box, and archives).
1h. Page footer (showing blog "links", search box, and archives).
|-
!2009-feb-01
|5 hours -> 40
| '''List of posts''', [http://www.fovea.cc/goodlooking/?cat=5 events] and [http://www.fovea.cc/goodlooking/?cat=3 news] pages (and more)
1h30. Cleaning the code. Extracting parts of the frontpage to external "template" files. I'm reusing the "news" box to create the "list-of-posts" page.
1h30. Hacking the list of items:
* making an header title bar (re-using front-page yellow-box style).
* Display only the thumbnail image of posts.
* Tweak specific styling for news, and tweak again.
2h. Specific styling for events:
* Creation of calendar image.
* Hacked in "event_calendar" plugin (was adding date to the post excerpt, which we do not care about)
* Work on the "event-date" box.
|-
|-
!Total
!Total

Revision as of 19:02, 1 February 2010

To Do list

<issues project = "Jeko" search = "false" filter = "false"/>

Discussions

Jeko's World Simulator Forum

Software Wars - French Edition

Work in progress

TorcsOgre WIP

OpenRacing WIP

OpenRacing

Using blender as a track editor

- Software Wars :: iTunes

- split RoR component

- RoR running on GMA 965

- Python AI for RoR

- AI_Traffic in Python

Daily journal

The time I spent working on vision project.

Period 1

Date Time Tasks
before 4 hours It is an approximation. I think it is honest, I hope that Keith feel that I am to.
2008-jun-09 5 hours Mails about scripting languages, about world simulator, writing planning, ...
2008-jun-12 4 hours Compile Torcs from CVS. Make it work (black screen problem). Start understanding the 3D engine and make improvements plans.
2008-jun-13 2 hours E-Mail. Understanding the 3D engine (car models and tracks). Make improvements plans.
2008-jun-24 2 hours Improvements plans, net-meeting with Keith about the world simulator framework.
2008-jun-26 2 hours Organize wiki. Answer a few questions. Installed and compiled mercurial's repository on my laptop.
TOTAL FOR JUNE 19 hours
2008-jul-02 4 hours Study OgreODE, Plans for the changes to an Ogre/ODE/Collada architecture., and share my point of view (e-mail meeting).
2008-jul-03 2 hours Install back Torcs / Mercurial after system upgrade. Fixing compilation problems.
2008-jul-04 3 hours Slimming torcs' data directory to 28 MB (te best that can be achieved without changing texture, sound and 3D models storage format. Birthday :-)
2008-jul-05 3 hours Trying hard to install OgreDotNet, I discovered it's a dead project supporting a very outdated version of Ogre... Trying alternatives. : Axiom, OsgDotNet, Irrlicht, Plib?
2008-jul-07 7 hours Ogre and OgreDotNet installed from sources.

Setting up my first Mono Develop Solution which links with OgreDotNet. Starting TorcsDotNet :

  • The track plugin is built as a library (containing all necessary code from Torcs to).
  • The C++ plugin is loaded from C# and initialized.
2008-jul-10 4 hours Learning Ogre3D
2008-jul-12 4 hours C# graphic plugin for Torcs.
2008-jul-13 2 hours C# wrapper for Torcs made a separate project from driving robot.
2008-jul-15 2 hours Torcs' C# graphic plugin added.
2008-jul-16 3 hours Torcs' C# graphic plugin : OgreDotNet initialized.
2008-jul-18 3 hours Torcs' C# graphic plugin :
  • Learning to marshal ansi strings.
  • Trying to convert street-4.acc file to Ogre3D internal format and display it. However even AC3D can't load it properly (and Blender to, logically)
2008-jul-19 4 hours Work on Torcs' C# interface and graphic plugin.
2008-jul-21 2 hours Cleaning up C# code.
2008-jul-28 0 hours Reinstalled everything.
2008-jul-29 4 hours C# plugins now displays something. I finally discovered (after hours of debugging) that custom window creation interferes with resources loading, which was the reason I got strange exceptions.
2008-jul-31 2 hours AC3D track file converted to Ogre Mesh, experimentally displayed with buggy textures.
TOTAL FOR JULY 49 hours
2008-aug-01 2 hours Fixed the texturing problems, switched texture format to JPG and PNG.
2008-aug-10 4 hours Fixed AC3D->Ogre conversion problems, work on the alpha channel problem.
2008-aug-11 3 hours Fixed double faced faces problem (have to be made by hand unfortunately), test of texture-based shadows.
2008-aug-20 4 hours Fixed alpha-channel textures, try to make shadows which work with this.
2008-aug-21 4 hours Solved texture shadow problem, initial work on car display.
2008-aug-22 3 hours Work on car display, fixed the aspect ratio bug, big commit to mercurial.
2008-aug-23 4 hours Continue with cars management (now called Avatars for generalization purposes). Switch to shader based materials.
2008-aug-25 2 hours Continue with cars management, discovered the Ogre picture is not dynamic (SSG steal it the OpenGL context after the 1st frame)
2008-aug-26 3 hours Investigate the Ogre + SSG OpenGL Context problem... No solution for the moment, posted questions on forums.
2008-aug-27 6 hours Solved the OpenGL Context problem. Fixed car position tracking correctly (there was a misalignment in C->C# data structures).
2008-aug-31 2 hours Work on the car orientation problem.
TOTAL FOR AUGUST 37 hours
2008-sept-5 2 hours Solved the car orientation problem. Camera can now track a car.
2008-sept-8 4 hours Fixed stencil shadow problem (by spliting the models), activate per-vertex ambiant coef, imported an high-quality skybox, start making some 3D models manifold.
2008-sept-10 4 hours Trying to add lightmaps to the track model.
TOTAL FOR SEPTEMBER 10 hours
2008-oct-5 2 hours Looking for OgreDotNet render-to-texture way of doing screenshots for multiple cameras.
2008-oct-6 3 hours Conception and start implementing the screenshot feature.
2008-oct-8 5 hours Code clean-up and refactoring, continue with the screenshot feature.
2008-oct-9 3 hours Manage the list of user defined cameras. [Keith paid for all hours (128) through now * 2]
TOTAL FOR OCTOBER (1) 13 hours

TOTAL FOR PERIOD 1 : 128 hours [PAID]

Period 2

Date Time Tasks
2008-oct-21 3 hours Configure and generate Doxygen for Monographic. Continue the camera stuff and some code re-arangement.
2008-nov-24 2 hours Getting back to the code, end of refactoring.
2008-dec-01 2 hours Commit to mercurial, try to fix the hack who force the 406 to be loaded (can't be fixed until we're more advanced, since it require the whole drivers XML files to be parsed).

Render the user-requested cameras to files (for easy sharing point of views requested by the vision engine).

2008-dec-02 4 hours Solved a Monodevelop bug (I wasn't able to compile anymore...). Work on "TorcsDotNet", the Track is loaded using the C++ plugin from a C# application.
2008-dec-04 4 hours Work on "TorcsDotNet", the Ogre graphic module is initialized and displays the Track, start investigating the creation of a tSituation (found raceman.h / raceinit.cpp).
2008-dec-08 3 hours Work on "TorcsDotNet", wrapping the raceengine for creation of drivers and situation.
2008-dec-09 2 hours Wanted to know what robots exactly need to drive (C version and SharpyCS one). SharpyCS uses nothing different than C robots (ie torcs C interface).

Decision is to wrap the C robots so we'll have all drivers available.

2008-dec-10 2 hours Work to wrap raceinit's InitCars, which initialize olethros. It bugs... Trying to figure out why.
2008-dec-13 2 hours Compile Olethros within Monodevelop project. Wrapped GfParm to load torcs' parameters files from C# RaceEngine.
2008-dec-15 3 hours Olethros loaded and initialized. Now we go to another big part : initialize a race.
2008-dec-16 4 hours Better wrapping of a few core classes. Start loading the car structure (tCarElt).

Study of Torcs source code to understand all the informations available/needed for cars simulation and driving.

2008-dec-17 3 hours Including SimuV2 and SOLID-2.0 into TorcsDotNet. Fighting against a Monodevelop bug...

Trying to figure out how I can monitor my activity inside Monodevelop (for precise time management).

2008-dec-18 4 hours Starting to wrap in C# SimuV2

Fixed all C# <-> C++ structure alignment bugs using __attribute__((__packed__)) on C++ side and [StructLayout (LayoutKind.Sequential, Pack=1)] on C# side...

It crashed.. Hardcore debugging.. Finally discovered a bug in Mono!

Under some circonstances C# Packed structures are not duplicated correctly (miss a few bytes at the end). I need to reproduce it on a simple example to submit a bug report.

Removed the dependency over PLIB by including in the project the very few functions used by Torcs core.

2008-dec-19 1 hour Reproduce the mono bug on a small example, filling bug report: https://bugzilla.novell.com/show_bug.cgi?id=460608
2008-dec-20 4 hours My bug is said to be solved in mono's SVN, so I got it and am compiling mcs from source to check this.

Writing unit tests to make sure all C# <-> C++ data structure conversions are bugless (since those bugs are *very* hard to track, as I made the experience).

Adding a check for the 'Pack=1' bug: which warns if it occurs. Latest version of mono do not have this bug so I'm now gonna change my all source code... just presume users have a non-buggy compiler.

2008-dec-23 2 hours Install/compile Torcs, Mono, Monodevelop and OgreDotNet on my eeePC. Solve Torcs' GCC 4.3 compile errors.
2008-jan-8 3 hours Work on C# Simu wrapper. Initialize the CarElt array of tSituation. Thinking about C#/C++ data synchronization for dynamic structures.
2008-jan-12 3 hours Initialize TCarElt fields (by porting the raceinit.cpp code to C#). Facing a strange exception thrown by the mono runtime (the JIT precisely). I suspect it doesn't like my unmanaged memory allocations.
2008-jan-13 2 hours Investigate proper ways to implement C#/C++ interoperability. Reading MSDN and papers.
2008-jan-14 3 hours Investigate SWIG. Torcs C interface described. SWIG generated code integrated into the project. Making previous version (TorcsItf) obsolete.
2008-jan-15 3 hours Make nice SWIG wrapper for Torcs interface, clean existing code accordingly.
2008-jan-16 2 hours Improve Robot and Car wrapper, learn more deeply SWIG in order to extend its generated interfaces..
2008-jan-17 3 hours Improve Track, ModInfo and Situation. Create an object oriented version of GfParm using SWIG. Rename variables so they starts with uppercase in C#.
2008-jan-20 4 hours Fix GfParm wrapper, learned own to customize SWIG generated code.
2008-jan-21 3 hours Finish to fix all compilation bugs related to the changes in the interface, so we're now back to previous state but with a much cleaner code.

Getting back to initialization of car structures (tCarElt) -- which doesn't segfault anymore (yeepee!).

2008-jan-22 4 hours Apply some polish to Car and Track interfaces. Write the C# version of InitStartingLine(). Move some code from Main() to the RaceManager.
2008-jan-23 4 hours End of the initialization stuffs. The simulation run, car positions are updated... but get random values.

Some important structure members might be uninitialized. Need to do some debugging.

2008-jan-24 7 hours Completed SwigTorcs. Learned how to wrap array structure members to C# array. Wrote lots of Swig Macros.

Been through all torcs interface headers to rename all variables and wrap all arrays cleanly.

Splitted Swig wrapper into several files: simulator/interface/*.i

Wrote a script which compiles the code, so I do not always need to launch Monodevelop (compile.sh at torcsdotnet's root).

2008-jan-26 2 hours C# simulation manager reached a working state. Plugins are initialized correctly, the car runs smoothly around the track.
2008-jan-27 4 hours Registered the project on sharesource.org. Preparing a package which compiles with autotools, check for dependencies and everything (in progress).
2008-jan-28 5 hours Projet accepted by sharesource.org. Packaging source with autotools complete (I need to package data now).

Removed some hard coded path in the graphic engine code by reading appropriate variables. Improved OgreDotNet renderer's management of resources (textures and stuffs).

2008-feb-01 4 hours Commited project to ShareSource. Adding little document on project wiki for installation. Fixing installation bugs.

Cleaning the data folder (3D models are now stored in zip files together with their textures and shaders).

Installing user-mode-linux in order to test compilation/installation. On various systems and from an empty system, to write a clean "how-to-install" documentation.

2008-feb-02 7 hours Finishing to clean every hacks. Now ./configure && make && make install do everything for you!

Updating "how-to-install". Preparing the interface for C# robots. Creating a stupid SharpyRobot.cs which drives straight to the wall as an example.

If WD like to ports Sharpy to OpenRacing, I'll start writing minimal GUIs for selecting Robots and Tracks, or start working on user inputs.

2008-feb-03 6 hours Exploring GUI libraries. Many things available for Ogre but just a few have C# interface, and most them are dead projects...

I got MyGUI working, the example is nice looking and the API simple so I wrote minimalist C# wrapper for basic functions.

I'm now trying to have something working with this inside OpenRacing.

2008-feb-04 6 hours I improve MyGUI# (I found my C# DLL wasn't working), now it's clean! Continue with the wrapper (as well as learning how MyGUI works).

Created an example OgreDotNet + MyGUI# application for demo and experimentation purposes.

OpenRacing now initialise MyGUI and add widgets on screen (just useless buttons, images and windows for now).

Some conception for integrating GUIs in OpenRacing had to be made: the graphic plugin is now in charge of rendering GUIs.

(because as we seen in Torcs, rendering of GUIs and 3D stuffs inside the same window with two different engines is not a good idea)

2008-feb-06 3 hours GraphicModule can now be initialized without being given Track and Cars, so we can load only the GUI. Starting input support using OIS through SWIG's framework.
2008-feb-08 3 hours Initial input support is working, wrapper dispatches EXIT events. Contacting MyGUI's guy to see if they are interested by a SWIG wrapper.

(problem is that they've already written a Managed C++ wrapper but Managed C++ isn't supported by Mono).

2008-feb-10 3 hours Minimal GUI is now working (with mouse and keyboard, using SWIG directors to specify C# callbacks to a C++ API).

Analysing torcs-ng mailing list. Working on the email for Torcs-NG guys.

TOTAL FOR PERIOD 2 : 128 hours [PAID]

Period 3

Date Time Tasks
2008-feb-11 4 hours (4h) Finish GUI. Clean-up scene after a race (so a new one can be launched). Sent the mail to Torcs-NG (and Torcs). Chat with Keith.
2008-feb-12 3 hours (7h) Working on Torcs-NG installation manual, trying to "de-hack" the GUI wrapper so I depend on MyGUI's SVN unmodified. Moving the minimalist SWIG wrapper for MyGUI within OpenRacing.
2008-feb-14 3 hours (10h) Sending mails (and brainstorming about it) to Torcs-NG. Installing OpenRacing and dependencies from scratch to be sure the installation manual will be good.
2008-feb-17 4 hours (14h) Porting GUI code to MyGUI's SVN, and debug (Strange exception thrown)...
2008-feb-18 4 hours (18h) E-mails... Sync my dev with my work branch on launchpad. Getting to know more about Torcs-NG human driver code. Debugging (MyGUI initialization still fail miserably).
2008-feb-19 6 hours (24h) Implements camera following car. Add command line options to disable GUI and graphics.

Compile and install torcs-ng. Studying trackgen-ng.

Tracking MyGUI's initialization bug. I give up until the official wrapper is out.

Ogre display suddenly became buggy too. It's probably due to the fact I installed ogre 1.4.5 (from ubuntu source package) to replace 1.4.9.

Updated MyGUI's SVN (to see if there is anything new for C#). It didn't compile anymore... Fixing a few stuffs.

Add few blueprints to launchpad..

2008-feb-20 3 hours (27h) Start refactoring Trackgen, understanding current design and thinking of next-gen one. Code pushed to a separated bzr repository:

https://code.launchpad.net/~jeko-ios-software/openracing/trackgen2

2008-feb-23 3 hours (30h) Mails mails mails.
2008-feb-25 3 hours (33h) Get ogre 1.4.9 back. Graphics are back, so it's true the graphics problems came from this. Commit and push to bzr. Mails & chat with Kilo and Keith.
2008-feb-28 5 hours (38h) Chat, mails and wiki about UTI's design.

Reverted GUI to a forked version of MyGUI containing SWIG C# wrapper. Created a bazaar repository.

Solved a bug for car orientation (discovered this one thanks to the car chasing camera).

2008-mar-10 5 hours (43h) Help Imre with the weird bug. Solved MyGUI's crash. Initial handling of OIS from OpenRacing, creation of the Human robot module.
2008-mar-11 4 hours (47h) Working keyboard driving with OIS (yeah it's cool!). Implementing Joystick support on OIS side, but it's not handled by human robot yet.

And it's untested since I don't have a joystick (I'm going buy one right now).

2008-mar-12 4 hours (51h) Implementing joystick integration... But it doesn't work. Impossible to get to joystick recognized and work with OIS.

OIS have to be launched as root. Then joystick is recognized, but not working... Finally upgrading to OIS version 1.2.

2008-mar-13 4 hours (55h) Finish joystick support. Some planning with Keith and Imre. Helping Imre with debugging.
2008-mar-14 3 hours (58h) HUD for RPM and speed.
2008-mar-16 5 hours (63h) Download and test Imre's PhysX branch (to get an idea how physx engine can be integrated into OR). Fixing some of its hacks.

Help Imre with debugging.. Study Torcs' OpenAL code (O.R misses sound, and I thing I can SWIG-wrap this code from Torcs quite easily).

Initial ODE integration (Tao's libraries are a joy to use!), learn how ODE works.

2008-mar-17 7 hours (70h) Work with Imre on ODE integration. Free-ride OpenRacing almost working.
2008-mar-18 3 hours (73h) Debugging free-ride on simuv2's side with a custom raycaster, I was able to do the first free ride (ended by a crash).
2008-mar-19 2 hours (75h) Give a hand to Keith for OR installation. Explore ODE bug.
2008-mar-20 2 hours (77h) Found ODE bug: 64bits machins requires to upgrade to ODE 0.11. Fixing Tao accordingly.
2008-mar-23 2 hours (79h) Adding an option to turn shadows off. Add an XML configuration file.
2008-mar-24 3 hours (82h) Adds a configuration GUI, store conf to XML, allows dynamic change of rendering quality, cleanup GUI code.
2008-apr-03 4 hours (86h) Installing Jaunty, installing OpenRacing on it, making the initial monodevelop 2 project.

TOTAL FOR PERIOD 3 : 86 hours [PAID]

Period 4

Date Time Tasks
2008-apr-22 3 hours (3h) Reading emails sent during last weeks. Getting monodevelop 2 knows about assembly references (for autocompletion and debugging).
2008-apr-23 2 hours (5h) Got better understanding of Makefile integration, had to recreate a Monodevelop project from scratch.

"Gac" references now can be found and are synchronized with makefiles, but the project doesn't compile or execute.

2008-apr-24 2 hours (7h) Turning in round, can't find how to make those no-in-gac DLLs to be found by monodevelop, which returns me nothing but cryptic exceptions while trying to compile.
2008-apr-27 3 hours (10h) Finally done with it. It requires some little hacks to make everything work, but at least that does the job. Merged and commited to trunk. Wrote howto on the wiki.

I really hope this will keep working, coz' this wasn't a funny task! :-P

2008-may-3 2 hours (12h) Trying to compile Imre's "cleaned" branch. Cleaning namespaces of trunk.
2008-may-4 2 hours (14h) OpenAl initial integration. Successfully emulated sound of engine with pitch and volume.
2008-may-6 2 hours (16h) Design of the AudioManager interface. OpenAL implementation of an AudioManager (not taking into account positions of other avatars for now).

It's enough for you to hear the great sound of your engine! (gives life to the car)

2008-may-9 1 hour (17h) Cleaning a little the AlAudioManager. Valgrinding the application hunting for badly unallocated memory. Trying to have clean init / shutdown and better handling of exceptions.
2008-may-10 7 hours (24h) Cleaned and improved modules (mostly graphics) initialization and shutdown process. Datas are cleaned properly when leaving a race.

http://bazaar.launchpad.net/~jeko-ios-software/openracing/work/revision/17

2008-may-13 3 hours (27h) MyGUI doesn't re-initialize well: ie GUI is invisible, exploring the code to understand what happens.

Side task: trying to compile MyGUI's current SVN repo, which contains a pure C# wrapper. Unfortunately, the SVN doesn't compile...

I fixed the first 3 compilation errors, all 3 coming from non-standard implementation of STL by Visual.

The next compilation error being a big job, I will suggest them at least to tag last version which compiles under GCC (still hoping they will fix everything by themselves).

2008-may-17 3 hours (30h) Trying to fix Ogre shutdown, trying to compile MyGUI's svn repo...
2008-may-18 2 hours (32h) Fixed Ogre and MyGUI's shutdown and init.
2008-may-23-24-25 5 hours (37h) Cleaning the disk + backup + installing Ubuntu 9.04.

Some issues:

- I tried to install it on an ext4 file-system, I hard hard time figuring why it couldn't boot. Simply because grub can't handle ext4... (so why the hell is this an option?)

- Following installations failed (tried 3 times).. hopefully I could finally manually finish the installation (so why the hell do I still use installers? ;-)

2008-may-26 2 hours (39h) Installing Open Racing and all dependencies.
2008-may-30 3 hours (42h) Compiling Imre's "cleaned" branch and study of the code, specifically his new ODE <-> simuv2 interface.
2008-jun-03 3 hours (45h) Merging some "cleaned" branch features into trunk: simuv2 without collisions, better ODE ray casting.
2008-jun-04 3 hours (48h) Finish merging "cleaned" branch, perfect free-ride without using Ogre file format as only input.

Focusing at understanding the Collada 3D format design, blender Collada plugin, and how to create the track edition tool-chain.

Current idea is to extend the Collada plugin to store additional informations to the track 3D model into <extra> nodes...

Open question now is how to specify them from blender.

2008-jun-05 6 hours (54h) Reading of Collada 1.4 specification. First attempt to create simple track with starting positions and lanes into blender.

Studying blender python API, starting custom XML exporter (eventually) compatible with Collada, adding road/track/lanes informations to the file format.

2008-jun-06 3 hours (57h) Work on blender modeling and blender scene exporter.
2008-jun-08 2 hours (59h) Work on the scene exporter. Defining a generalist file format to describe a track (using only primitives blender can create).
2008-jun-10 2 hours (61h) Trying to find path vertex radius getter in the API (but it seems like it's not exported)... Thinking about roads and lane specifications.
2008-jun-14 2 hours (63h) Finalize the test track, thinking about UTI.

Conceived an algorithm for automatic lane width computation, which solves the "vertex radius not exported problem", and will makes track creation process easier.

2008-jun-15 1 hour (64h) Blender.
2008-jun-21 1 hours (65h) Thinking about UTI.
2008-jun-22 2 hours (67h) Thinking about UTI, setting up initial plans to simplify c++ code dependencies (to have more freedom with UTI).
2008-jun-23 2 hours (69h) Killing some C++.
2008-jun-24 1 hour (70h) Killing some C++. The important step is that I successfully removed simuv2's dependency over the track format. By cascading effect, we will eventually get rid of torcs' track plugin.
2008-jun-25 1 hour (71h) Recruiting. Starting to kill the track plugin.
2008-jun-26 3 hours (74h) Removed track, plib and robottools, raceman from libsimulator. The C++ code doesn't depend any more on the track format. Writing report, planning next phase.
2008-jul-15 2 hours (76h) Conception of UTI, hand-drawing a global view of UTI's interactions with other modules: Ode, Ogre, Robots and Blender.

Defining the PhysicalScene in C#. Study of OgreMesh format for the .OTF to OGRE conversion tool.

2008-jul-16 1 hour (77h) Continue with conception of UTI, sending email to Miguel.
2008-jul-20 2 hours (79h) Defining UTI.cs, documenting here: Achieving UTI.
2008-jul-21 1 hour (80h) Continue with UTI, extending the architecture to a (in my opinion) cleaner 2 layers model.
2008-jul-22 1 hour (81h) Fixing autogen bug, fixing compilation doc. UTD.
2008-aug-01 3 hours (84h) Finishing UTI and UTD specification. Starting an preliminary implementation of it, creating big mess in Ode and Ogre modules...
2008-aug-03 2 hours (86h) Ode imports its data from UTD. Command line preprocessing options for UTD to Ogre conversion.

TOTAL FOR PERIOD 4 : 86 hours [86 hours PAID]

Period 5

Date Time Tasks
2009-aug 3 hours (3h) (Aggregate of small tasks done during august)

Some OpenRacing UTI coding, e-meeting with Keith, downloading RoR, reading forums and source code of RoR. Thinking about our goal in RoR. AI?

2009-sept-01 1 hour (4h) Reading RoR source. Trying to compile RoR.
2009-sept-.. 4 hours (8h) RoR and dependencies downloaded and compiled, read source code, testing RoR, thinking and planning, blogging about and graph.
2009-sept-17 1 hour (9h) Email with Keith, reading the forum about current work RoR.
2009-oct-1 3 hours (12h) Updated RoR from svn. Get it compiled and working again (it takes long, I should ask/find how are developers doing).

Updated code is very buggy, the cars slips on the road. Looking into forum if I'm the only one with the problem.

Registered to the forum. E-mailed Imre about his tasks on RoR... and if he thinks I can help. Reading the work-in-progress list, to get a picture of what they are doing now.

2009-oct-3 3h (15h) Talking with RoR team, got introduced by Imre on the IRC. Very productive talk, I've got better overview of overall architecture and RoR's current direction. They are very open, I like the admins :)

Reading this and this.

Updated this

2009-oct-4 3h (18h) Participating to the monthly Sunday meeting.

I had the opportunity to present myself and our objectives to RoR team, they are welcoming us and enthusiastic about self-driving cars project.

They believe it makes sense to write the AI in higher-level language than C++. They are open to integrating Python binding inside RoR.

2009-oct-5/6 2h (20h) Start documenting the code: about initialization stuff for now, I will propose some cleaning. I got a patch pushed in about the python building script, nothing much but it's good start.
2009-oct-7 2h (22h) Writing this.
2009-oct-12 1h (23h) Understanding RoR's general terrain/roads/world management... Which I started to explain here.
2009-oct-15 4h (27h) Reading the oct-12 meeting (which I missed). Answering to interesting threads: pushing for some python, code cleaning, global design of the application, separation of game mode and simulation mode, better project management (no clear milestone are defined, ongoing features are managed as bugs, etc.)

TOTAL FOR PERIOD 5 : 27 hours

Goom

Warning: It's a low-priority task.

Tasks

This are tasks suggested by me, by keith or by users. I'm making a raw list and will apply filters later.

  • Report what is necessary from the following to Sourceforge's TODO list
  • Proceed patch from Séven Le Mesle, 10-jan-2005 (slap me!).
    • Which adds support for gtk-2 and provide a Beep Media Player (BMP) plugin.
  • Proceed patch from Chris Seaton, 23-feb-2005 (slap me again!)
    • Which compiles goom as a DLL for Windows from Visual Studio (to make integration into new media player easy).
    • [I sent him an email to know of possible updates]
  • Turn the "big goom logo" into something nicer and less controversial.
  • Put my stuff into Banshee
  • Improve totem integration
  • Support 16:10 and 16:9 screens under WMP
  • Get back in touch with the libvisual team
  • Goom for iTunes windows
  • Goom for foo2000
  • Proceed patch from Stanislav Brabec, 4-apr-2005 (you know what to do..)
    • Which fixes warning under GCC 4
  • Ben davis planned to write a ruby binding (may-2006), ask for news.
  • OpenGL based Displacement Filter, Experiment the tutorial by Sylvain Levebvre.
  • C# goom or C# binding for goom
  • Proceed patch from Rémi Duraffort, 24-mar-2009

Daily Journal

Time I spent on Goom project.

Date Time Tasks
2008-dec-01 1 hour Install dependencies needed to compile Totem.
2009-apr-24 X hour Try to retrieve stuffs-to-do from all over internet and outer space.
Total 1 hour

Goom

Software Wars eBook

For iPhone and Kindle.

Time Tracking

Date Time Tasks
2009-nov-16 0 hours Setup an iPhone project. Display a fullscreen rich text view and a button which flip to a menu.

Save scrolling position when leaving, restore it when re-launching the application.

2009-dec-4 5 hours Export to HTML, automatize cleaning of the generated file with some shell/sed/vim scripts.
2009-dec-7 2 hours Finish the cleaning (broken table of content and footers), test on Amazon's Digital Text Platform, final touch.
2009-dec-8 1 hour Last fixes, upload and test final version on DTP, send it to Keith, archive last version of the tools.
Total 8 hours

Mini Spec for iPhone App

  • First launch shows the forewords (probably we need to add a "Welcome the iPhone version of Software Wars" blah blah)
  • On next launches, we restore the application to the position in the eBook it had when leaving it.
  • When reading, a little button allows to go to the menu.
    • To move to another chapter
    • To change font size (small / medium / large)

Mini Planning

  • Dev: 10 hours [2 hours done, probably around 2 remaining]
  • Export the book text to iPhone friendly format: ~10 hours [DONE]
  • Integrate (and resize) some (or all) images into the eBook: ~10 hours (if just a selection of images) [DONE]
  • Clean the layout. [DONE]

(I'm not sure if I'm optimistic or pessimistic with those numbers).

-- I was pessimistic.

g...l.......org

Work for GL

Time Tracking

Date Time Tasks
2009-jan-14 8 hours Setup the project (Install / BZR / ...)

Search for the appropriate starting theme

Install it, Start customizing.

Integrate properly the Searchbox.

Look for "Featured Content" plugins, which should be a very convenient way to automatically generate the front page from WP posts. (found nothing suitable for now)

2009-jan-15 4 hours -> 12 Found and installed a "post thumbnail maker". It is used it to specify an image for the front page "3-latests-news" box (news).

Completed the news box (still need multi-browsers tests), it's almost pixel-perfect on chrome and firefox.

Prepared the events box (for the upcoming events).

Initial full-page-height yellow border. There is sometime 1 pixel error (horizontally, so very noticable). I figured out why, but have to implement a good solution.

2009-jan-19 7 hours -> 19 Morning. 2h. Setting development environment in my home computer. Cloned the files + database. Wrote "clone.sh" script which automatizes database cloning. Wrote some documentation. Then I cloned on the laptop, it took only 5 minutes to get a full working copy of the website. I can now work from everywhere!

Afternoon A. 2h. Turned the Homepage to a WordPress static page. Content of the static page will go to the top "275px" box. All following boxes will be filled by hand made loops. (3 latest news, events, etc.)

Afternoon B. 3h. Got the top box filled with content (image + video + box with link to glTV). Found, installed and configured the "Smart Youtube" plugin to get the top video displayed. CSS Styling for everything. Commit&push. Update dev.fovea.cc/goodlooking database with new content. Documentation.

2009-jan-20 4 hours -> 23 Task 1. 2h. Trying to reproduce keith's youtube video layout problems. Blindly fixing many possible problems by making the site highly font-size "insensitive". Adding the yellow triangle next to "View more videos..." (was harder than expected, because using standard <ul><li> broke the layout on Chrome -- so maybe others).

Task 2. 2h. Create the 3 yellow boxes at the bottom. I made it an extensible way, by allowing boxes to have any width (it's a 1x28 image horizontally streched + the boy's head on top-left). It makes the page lighter and it will certainly be of use for other pages. I also took some time to look for the best font for block headers (i.e. the nearest from the JPEG). I'm using Helvetica bold 11pt. (using a text instead of image is lot better for SEO). Replaced the "image text" by a real text on "upcoming events".

2009-jan-25 4 hours -> 27 Fixing layout problems.

1h30 Fixing various Internet Explorer margins by playing with an IE specific CSS. Cleaner layout works on all version from IE 5 (tested OK with http://ipinfo.info/netrenderer)

1h30 Fixing the problems for "screen width <= 1034px". I first implemented an hacky solution with a Javascript code which checked the viewport size and adapted the layout accordingly. It was quite ok but 1px horizontal offset appears on even (non-odd) size. I finally found a much much cleaner and HTML-only solution by wrapping the page into two additional levels of "div": cross-browser and bug free!

30 min. Making the background image "screen width > 1280px" proof (by adding a gradient at borders)

30 min. Setup a public beta version of the site (http://www.fovea.cc/goodlooking). Test it on all browsers of known universe: http://browsershots.org

2009-jan-26 5 hours -> 32 Events and Twitter 1/2.

3h30. Integration of event-calendar. Posts can now have an associated "start-date" and "end-date". If those dates are set, the post will be considered as event. It will be made part of the event category and be listed into the 'upcoming event' box (which shows the 10 events to come).

1h30. Integration of latest Twitter messages into a box using twitter-for-wordpress.

2009-jan-27 5 hours -> 37 Twitter 2/2, Flickr and Database Administration.

1h. Fix latest problems with twitter integration.

3h Integrate Flickr, using tantan-flickr. Main problem is that most Flikr plugins are only able to display users pictures. I hacked it to make it display random images from the "bassndrum" flickr group. If this is an Ok solution, we have to create a LTJ Bukem specific group. Wrote CSS, PHP, usual stuffs...

1h. Wrote some scripts allowing to retrieve only parts of the database (content, options, users). Each part is pulled into a separated SQL file, which is cleaned from install-specific informations (replaced by tokens: @@@THE_URL@@@, @@@THE_PATH@@@). Those files are then managed by GIT, so we have an history of the database (no more fear when testing new plugins)

Why into separate files? Because we may like to replace "options" without erasing the "content = posts+comments" of an install.

2009-jan-29 3 hours -> 40 Facebook box and page footer

2h. Integrate and stylize the Facebook "become a fan" box.

1h. Page footer (showing blog "links", search box, and archives).

2009-feb-01 5 hours -> 40 List of posts, events and news pages (and more)

1h30. Cleaning the code. Extracting parts of the frontpage to external "template" files. I'm reusing the "news" box to create the "list-of-posts" page.

1h30. Hacking the list of items:

  • making an header title bar (re-using front-page yellow-box style).
  • Display only the thumbnail image of posts.
  • Tweak specific styling for news, and tweak again.

2h. Specific styling for events:

  • Creation of calendar image.
  • Hacked in "event_calendar" plugin (was adding date to the post excerpt, which we do not care about)
  • Work on the "event-date" box.
Total 40 hours