110 likes | 248 Views
CS -Status. Results from workshop 2007 what was planned for 2007 what was done in 2007 Packages Statistics Miscellaneous. CS Release 3.10 (short term), done no new features compared to CS 3.00 very few minor changes compared to CS 3.00 based on LV 8.20 ( LV 8.2.1 )
E N D
CS-Status Results from workshop 2007 what was planned for 2007 what was done in 2007 Packages Statistics Miscellaneous Dietrich Beck
CS Release 3.10 (short term), done no new features compared to CS 3.00 very few minor changes compared to CS 3.00 based on LV 8.20 (LV 8.2.1) using LV projects using a library (lvlib) for each class pre-alpha release in spring 2007 release in summer 2007 (August 23rd) CS Release > 3.10 (mid term), planned users do not wish major changes restrict to occasional maintenance releases (Spring 2008) CS Release > 4.00 (long term), no users do not wish major changes like switching to built-in OO features of LabVIEW Result from CS Workshop 2007, I Dietrich Beck
Device Base Classes. There exists an idea of defining base classes for device classes. Possible candidates are arbitrary function generator base class, power supply base class, oscilloscope base class. A few people like Holger, Dietrich and Frank will meet in February 2007 to discuss the requirements. done An alternative to base classes would be usage of template classes (example classes). no Other ideas. A class (or an extension of an existing class) that shows states (ERROR, READY, CONFIGURED, ...). done CSObjectNet. Have been presented by Holger and Alex and will be ready end of February 2007, done. Any volunteers for a first application? ? Result from CS Workshop 2007, II Dietrich Beck
Coding conventions. As an example, one could use the VI analyzer tool to ensure quality standards of code. Do we want that and who takes care of that? There is no commitment by the project leaders of the different groups to enforce such coding conventions. hm... Reusability. It would be nice, if one could establish standards for classes, so that they are less experiment specific and can be reused by others. Of course, this requires some work by the original developer. There is no commitment by the project leaders of the different groups. The only idea here is the usage of device base classes (see above). done or not The tool for the configuration data base needs some modifications. Who takes care of that? me CsDatabase. Some users have requested a versioned data base for settings. What are the requirements and who is willing to implement that? nobody We propose a next CS workshop on January 24-25 2008. almost Result from CS Workshop 2007, III Dietrich Beck
up to CS 2.40: huge 200 MByte zip file then: packages clear responsibilities for sub-components makes quality assurance possible http://wiki.gsi.de/cgi-bin/view/CSframework/CsBuildingPackages developer: take care only of my code user: know, what has been upgraded works well for main package works well for some device packages experiment specific stuff? Packager contents and dependencies defined in MS-Excel sheet packages contained in normal zip files automated creation of release notes automated building of packages with XML description Packages I Dietrich Beck
Unpackager dependency/conflict checking package checking installing packages removing packages downloading packages ... "three-click installation" Packages II Dietrich Beck
among top 1% of SourceForge projects > 800 downloads of CSMain 29 packages (of 2 maintainers) 42 (open 3) bugs 30 (open 1) feature requests 16 (open 2) patches 3 feature requests 8 developers Statistics from SourceForge Dietrich Beck
experiments requiring high flexibility experiments with a large variety of hardware types experiments with up to 10,000 (1M possible) process variables PHELIX SHIPTRAP ISOLTRAP LEBIT data taking Motion CaveA others ... Motion CaveA HITRAP commissioning REXTRAP FOPI development Usage of the CS framework today RISING ClusterTRAP LPT GSI, Germany Mainz, Germany Greifswald, Germany CERN, Switzerland MSU, USA Lanzhou, China TrigaTRAP Dietrich Beck
providing software to others is quite some work only software maintained by two (db, hb) people is available on SF idea of collaboration did not really work out experimentalists develop software under time pressure software too experiment specific (limited reusability) software is not officially released and available to others (SourceForge!) usage of CSAppicationBaseClasses (-> talk) documentation is... either too short (ten lines of text: incomplete) or too long ( 1 page of text: "let's call the developer") out of date at the wrong place really frustrating! features to include if requested: Shall we implement it? ("CS Access System") sometimes requested telepathically (like "Settings" for "Unpackager") Developers Point of View Dietrich Beck
SourceForge 2 main users + 2-3 occasional users bug reports, feature requests, monitoring, etc. used rarely maybe everybody uses the RSS feed of the "News"? Mailing List in total 14 messages in 18 months by one person () nine subscribers http://wiki.gsi.de/cgi-bin/view/CSframework/WebHome around 40 registered group members everybody should get notifications? What is your source of information? Information Flow Dietrich Beck
Compared to a one-step transition, the two-step transition from CS 2.40d4 to CS 3.10 in two steps did cost significant time. Maintenance (just cumulative patches) release CS 3.11 in spring 2008, based on LV 8.2.1. Switch to LV > 8.2.1 proposed after 2008. Subversion repository might be available outside GSI soon. Other Issues Dietrich Beck