On 8/10/07, Jack Schmidt <[EMAIL PROTECTED]> wrote: > > Something like that. Each test file contains a record of how long the > entire file takes in units proportional to how long a specific test > file takes. I believe the constant is currently chosen so that 1 unit > corresponds to 1msec on a 1ghz pentium class processor. > > The standard test suite runs the tests from fastest to slowest and > gives estimates on how long the test will take on the current computer > based on the recorded timings versus the actual timings of the > previous tests. > > This has proven adequate to detect several slowdowns in development > code before it was released and also provide a friendly progress meter > to those not using supercomputers.
Unfortunately a file-based approach won't work for SAGE, because the number of tests in any given file is (or rather -- should be!) constantly growing, hence it is normal or even good if files take longer to build than before. Also, tests can be reordered or slightly modified by just about anybody, which makes it hard to record the times they took, though that would be possible. I.e., I could run something that records the time on my computer to do each block of tests, and for ones that haven't changed, check that the timing hasn't gotten much worse. This could be reset every so often. This would have the massive advantage over my previous suggestion that it could be made totally automatic and would cover all code in SAGE. It has the disadvantage that when looking at docs users wouldn't have a clue how long they are supposed to take to run (since that information would presumably be in a central database for a given machine). -- William --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~----------~----~----~----~------~----~------~--~---
