Hi, This week I pulled out common runner code into runner_helper and graph code into graph_helper. I added a push runner and a route to the web interface for adding new push metric data. I modified pull_runner so that it only runs for pull metrics. I updated the tests due to the previous changes. I fixed the paths so that they are generated as absolute paths so that the code works no matter what directory it runs from. I added push_runner to the crontab file so that it runs to generate the graphs for push metrics.
I used autofmt_xdate() when graphing so that the xaxis labels are slanted so that they are readable. I fixed some of the manifest files where pull metrics were accidentally set to be push metrics. I removed the day locator from the xaxis of the graphs, because with the day locator there was a label for each day, which made the xaxis unreadable. Based on Piotr's suggestion, I used with open() in runner_helper since it automatically closes the file. I modified the makefile to use a variable for the manifest directory, because in some places it had a hard coded directory. I encoded the data that is POSTed to /push as ascii, because when I did not, the data was unicode which caused problems when I try to search for a matching key in the models dict. Finally I renamed the rc_bug metric to rc_bug_count to match the name of the table. In addition to those changes to the repo, I also did some work on push metrics on the server. I added a push script for rc_bugs_count[0] and added it to the crontab. It works and adds data every six hours, since that is how often the webpage being scraped is updated. For next week: I knew there had to be something I was forgetting. I forgot to make a list of what I will be working on next week. First of all, something is wrong with the releases and releases_count metrics. Releases is only displaying buzz with None as the releasedate and releaseversion. Releases_count was last updated June 10, so something is causing releases_count to rollback. So I will figure out what is wrong with those two metrics and correct it. Second, I want to use CSS for the tabs instead of JQuery UI, which is what is used now. Since the static interface that exists now is only going to be used by people with JS disabled, it makes sense to use some non-JS method of rendering tabs. Third, I want to have some way of automatically generating the index page of the web interface, so that when a new manifest is added a link to the metrics' page is automatically added. I was thinking of creating a manifest2index.py like manifest2orm. It could be run by the makefile, but it would take in all of the manifest files as parameters instead of being run on each individual manifest file. At that point, I think I am mostly done with the static web interface. So if I still have time, I will start on the dynamic interface. My idea is to have some JS that will replace the links on the index page with a dynamic graph. There will be a dropdown with a list of metrics and an "add metric" button. When a metric is selected and the button clicked, it will add the metric to the graph. There would be a list of metrics already added and red Xs to remove the metric from the graph. Also, my mentors and I will be debugging the test suite over IRC next week. I will make the graphs for push metrics only regenerate when the appropriate metric has data pushed. [0] - http://paste.debian.net/108141/ Joseph _______________________________________________ Soc-coordination mailing list [email protected] http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/soc-coordination
