Changes between Version 74 and Version 75 of DevProjects
- Timestamp:
- Mar 23, 2012, 10:26:15 PM (13 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
DevProjects
v74 v75 37 37 38 38 == Proposed features == 39 The following development projects are available '''(Note: please do not add items to these lists. 40 Suggest new items in the boinc_dev email list.)''''''' 39 The following development projects are available 40 '''(Note: please do not add items to these lists. 41 Suggest new items in the boinc_dev email list.)''' 41 42 42 43 === Web features === 43 Easy:44 44 45 *Use a WYSIWYG bbcode editor such as http://wysiwygbbcode.codeplex.com/releases/view/4983646 *Document the classes and IDs used by the PHP code; document the "standard" stylesheets, main.css and white.css.45 1. Use a WYSIWYG bbcode editor such as http://wysiwygbbcode.codeplex.com/releases/view/49836 46 1. Document the classes and IDs used by the PHP code; document the "standard" stylesheets, main.css and white.css. 47 47 Finish black.css, and maybe make another style. 48 * Combine user page and profile. #601 49 * Change default Q&A page to refer BOINC-specific questions to BOINC web site. #603 50 * Convert team name HTML and team description to BBcode. #604 51 * Add a Reset button to prefs (both general and project) 52 53 Medium: 54 55 * Add a mechanism where joining a team or group requires approval of an admin. #605 56 57 Difficult and/or complex: 58 59 * [wiki:PrefsRemodel Remodel the preferences system] 60 * Propagate profiles between projects. #608 61 * Same for forum preferences. 62 * Add new profile features: #609 48 1. Combine user page and profile. #601 49 1. Change default Q&A page to refer BOINC-specific questions to BOINC web site. #603 50 1. Convert team name HTML and team description to BBcode. #604 51 1. Add a Reset button to prefs (both general and project) 52 1. [wiki:PrefsRemodel Remodel the preferences system] 53 1. Add new profile features: #609 63 54 * list of recent posts and threads this person created, on this and other projects. 64 55 * other features from social networking sites? … … 67 58 List referrals on user page (show only those still active). 68 59 Add new referral table to DB. 69 * Make it easy for teams to offer a client download that features their skin,70 and pre-register the user on that team for any projects he attaches to. #61071 60 72 61 === Applications === 73 Easy:74 62 75 * Create Makefiles and project files to build the sample applications using MinGW and Dev-C++. 76 77 Medium: 78 79 * Write example FORTRAN application and Makefiles/ project files. 80 81 Difficult and/or complex: 82 83 * Investigate the crlibm library for generating identical results across processors 63 1. Create Makefiles and project files to build the sample applications using MinGW and Dev-C++. 64 1. Write example FORTRAN application and Makefiles/ project files. 65 1. Investigate the crlibm library for generating identical results across processors 84 66 (or at least reducing the number of cases for HR). 85 *Java support: core client checks for the existence of JVM, reports version to scheduler.67 1. Java support: core client checks for the existence of JVM, reports version to scheduler. 86 68 Write Java wrapper (runs JVM, gives it jar files). Note: Sztaki has already done some part of this. 87 *Same, .NET69 1. Same, .NET 88 70 89 71 === Client === 90 Medium:91 72 92 *Keep track of the statistics of how long it takes to upload files, and to report results.73 1. Keep track of the statistics of how long it takes to upload files, and to report results. 93 74 Use that info to improve compute deadlines (e.g., subtract the 2 sigma point for both). 94 *Add an "expiration delay" for files: delete them only after they haven't been used for X time.95 *Core client monitors total CPU usage, limits its own CPU usage so that total is < 100%.96 *Monitor working-set size of non-BOINC apps; don't allow total to exceed physical RAM.75 1. Add an "expiration delay" for files: delete them only after they haven't been used for X time. 76 1. Core client monitors total CPU usage, limits its own CPU usage so that total is < 100%. 77 1. Monitor working-set size of non-BOINC apps; don't allow total to exceed physical RAM. 97 78 This could eliminate the need for memory-usage preferences. 98 79 Note: this might be tricky because parts of a process's working set is shared with other processes. 99 80 Some study is needed to figure out the best approach. 100 * GUI RPC to tell apps to checkpoint and quit. 101 * Keep track of upload/download sizes, show weekly/monthly transfer totals. 102 103 Hard: 104 105 * Use multiple disks. 81 1. GUI RPC to tell apps to checkpoint and quit. 82 1. Keep track of upload/download sizes, show weekly/monthly transfer totals. 83 1. Use multiple disks. 106 84 This would increase available space, 107 85 and (if slot dirs spread across disks) would allow I/O-intensive apps 108 86 to run more efficiently. 109 87 Would have to remove assumption that project dir is ../../projects/X relative to slot. 110 *Have the core client sense CPU temperature and throttle CPU if it goes too high.88 1. Have the core client sense CPU temperature and throttle CPU if it goes too high. 111 89 [http://www.lm-sensors.org/ Open-source software for collecting sensor data (on Linux)] is available. 112 *Windows: get proxy config info directly from the OS.113 *After an applications exits or is killed (for whatever reason) make sure90 1. Windows: get proxy config info directly from the OS. 91 1. After an applications exits or is killed (for whatever reason) make sure 114 92 (after a few second delay) that its subprocesses are gone too. 115 93 Don't restart the job until this happens. 116 94 Unix: use process groups and killpg(). 117 *More generally: make a better state machine for shutting down apps:95 1. More generally: make a better state machine for shutting down apps: 118 96 tell them to checkpoint, wait a little, tell them to quit, clean up straggler processes. 119 *Same, but for suspend: if we tell a client to suspend and it's still97 1. Same, but for suspend: if we tell a client to suspend and it's still 120 98 using lots of CPU after a few seconds, abort it (or something). 121 *Integrate !BitTorrent ([http://www.rasterbar.com/products/libtorrent/ libtorrent]?)99 1. Integrate !BitTorrent ([http://www.rasterbar.com/products/libtorrent/ libtorrent]?) 122 100 with the core client (Janus Kristensen is working on this). 123 *Log result start/ends (for use by 3rd-party software like !BoincView).124 *Vista: if get 'about to shut down' msg from OS,101 1. Log result start/ends (for use by 3rd-party software like !BoincView). 102 1. Vista: if get 'about to shut down' msg from OS, 125 103 stop apps immediately (don't tell them to checkpoint). Investigate. 126 104 127 105 === Installer === 128 *Check if host is configured to go into Standby mode after X idle minutes.106 1. Check if host is configured to go into Standby mode after X idle minutes. 129 107 If so, ask user if they want to let BOINC run for X minutes, then go into standby right after that. 108 1. Make it easy for teams to offer a client download that features their skin, 109 and pre-register the user on that team for any projects he attaches to. #610 130 110 131 111 === BOINC Manager === 132 Easy:133 112 134 * Show credit numbers (especially totals) with thousands-groups separators (e.g., commas)135 * If using an AMS, put link to AMS (or user page) in sys tray popup, elsewhere.136 113 137 Hard: 138 139 *Reimplement the BOINC Manager so that it is:114 1. Show credit numbers (especially totals) with thousands-groups separators (e.g., commas) 115 1. If using an AMS, put link to AMS (or user page) in sys tray popup, elsewhere. 116 1. Reimplement the BOINC Manager so that it is: 140 117 * Configurable via an XML file 141 118 * Skinnable via CSS … … 143 120 144 121 === Server and Back End === 145 Easy:146 122 147 * Add a mechanism to recover from situations where the scheduler fails to send its reply message. 123 124 1. Add a mechanism to recover from situations where the scheduler fails to send its reply message. 148 125 * Add a reply_pending field to host table. Set at start of scheduler, clear after reply message sent. 149 126 * If get a request from host with reply_pending==1, scan for lost results. 150 127 This would replace the <resend_lost_results> mechanism. 151 128 NOTE: this is not a good design; successful send doesn't imply successful receipt. 152 * XML stats: add lat/long to user record? 153 154 Medium: 155 156 * Enforce file immutability. 129 1. XML stats: add lat/long to user record? 130 1. Enforce file immutability. 157 131 I.e., have create_work() check if any file is already in the download hierarchy, 158 132 and if so make sure that it is the same. 159 133 Use MD5s for efficiency. 160 161 Hard: 162 163 * Add support for MPI-type applications: try to schedule all the jobs in a batch more or less simultaneously. 134 1. Add support for MPI-type applications: try to schedule all the jobs in a batch more or less simultaneously. 164 135 NOTE: being done by Volpex project from U. of Houston. 165 *Implement a mechanism so that server software detects incompatible database format. (related to #715)166 *Add project_specific_info field to host table (text).136 1. Implement a mechanism so that server software detects incompatible database format. (related to #715) 137 1. Add project_specific_info field to host table (text). 167 138 This stores, for each app, the ID of last reference WU completed, 168 139 and the info returned by that WU (as part of its stderr output, in a <app_info> tag).
