History of jbidwatcher/trunk/src/com/jbidwatcher/app/JBidWatch.java

  • @694
    avatar 694

    Make sure there's an error logger set on startup.

    by mrs, 21 Dec, 2008 01:08 AM
  • @693
    avatar 693

    Finally clean up the previously deep tangle between JConfig and ErrorManagement. All error logging is now handled through JConfig.log(), so ErrorManagement doesn't need to be a static block of functions anymore. Much cleaner, imo, but a pretty far-reaching change.

    by mrs, 21 Dec, 2008 01:04 AM
  • @691
    avatar 691

    Fix the AuctionServerManager/AuctionServer tangle with a nicer and cleaner fallback architecture.

    by mrs, 21 Dec, 2008 12:05 AM
  • @689
    avatar 689

    Even more sweeping changes to fall back to a non-ebay.com eBay server (ebay.co.uk right now) for items which ebay.com is refusing to show.

    It appears to work, and logging in even appears to work(!), but it needs real world testing, and real world bidding, buying, and sniping.

    This is probably the largest set of deep modifications I've had to make in many years to support an eBay change. It may make certain features harder in the future, but it's necessary to make it work again right now.

    [#508]

    by mrs, 16 Dec, 2008 12:09 PM
  • @688
    avatar 688

    More sweeping changes to fall back to a non-ebay.com eBay server (ebay.co.uk right now) for items which ebay.com is refusing to show. Rudimentary tests show the core implementation is successful.

    by mrs, 09 Dec, 2008 12:24 PM
  • @684
    avatar 684

    Some preliminary refactorings, working towards enabling multiple simultaneous eBay servers, so I can have a server that resolves ebay.com and a server that resolves non-ebay.com items, and when a snipe fires or auction reloads it'll use the right one. This refactoring doesn't do that, but it makes it easier to do that.

    by mrs, 29 Nov, 2008 01:02 PM
  • @662
    avatar 662

    Perform login cookie updates at odd intervals, so as to not be as predictable.

    by mrs, 27 Oct, 2008 07:54 AM
  • @636
    avatar 636

    Try to simplify the database access, by making _everything_ save itself when it changes. No more automatic saves, queued saves, etc., since all that crap is dangerous with multiple threads doing simultaneous database access. Hopefully this will yield more consistent (and good!) behavior.

    by mrs, 15 Sep, 2008 09:50 AM
  • @634
    avatar 634

    A few thread-related cleanups, and try to load the scripting subsystem in an alternate thread so it doesn't block startup as much.

    by mrs, 15 Sep, 2008 01:12 AM
  • @615
    avatar 615

    Make the error log monitor work again.

    by mrs, 24 Aug, 2008 10:53 PM
  • @605
    avatar 605

    Don't try to load the auctions xml file if we didn't create the database this run. In other words, only the first time we load JBidwatcher as an upgrade should we try to import the auctions.

    by mrs, 14 Aug, 2008 05:54 PM
  • @579
    avatar 579

    Try to improve the USB portable support by rewriting all the config values with paths to point to the USB drive when '--usb' is passed.

    by mrs, 27 Jul, 2008 11:18 PM
  • @573
    avatar 573

    Preserve changes to the My eBay search on restart. We were overwriting it, because searches were loaded later than the auction server, and the auction server was deciding that there wasn't a 'My eBay' search created yet, and then when the searches loaded later, the one from the save file was being ignored, since it already existed.

    [#188 state:resolved]

    by mrs, 24 Jul, 2008 05:26 PM
  • @568
    avatar 568

    Clean up minor scripting detritus related to the upgrade to 1.1.3.

    by mrs, 24 Jul, 2008 12:04 AM
  • @561
    avatar 561

    Only load up one memory debug box when debug.memory is true, instead of two.

    by mrs, 14 Jul, 2008 06:05 AM
  • @547
    avatar 547

    Make uploading auctions to My JBidwatcher work.

    by mrs, 04 Jul, 2008 11:12 AM
  • @531
    avatar 531

    Finally rename JBidMouse (which is very inaccurate) to UserActions which describes the code in it much better.

    by mrs, 01 Jul, 2008 02:17 AM
  • @530
    avatar 530

    Finally eliminated a lot of the nasty tangles of code references (a references b references c references a) in the UI code. It's pretty decently clean now, and in the process the code has been factored out reasonably.

    by mrs, 01 Jul, 2008 02:15 AM
  • @529
    avatar 529

    More reduction in dependence on FilterManager. Split out the list-management properties of FilterManager, as well as the AuctionListHolder into a top-level class. Set it's properties in an initializer, and try to clean up as much as possible.

    by mrs, 30 Jun, 2008 09:33 AM
  • @528
    avatar 528

    More reduction in dependence on FilterManager. Slowly but surely I'm eliminating the nasty tangle at the heart of the program.

    by mrs, 30 Jun, 2008 08:27 AM
  • @527
    avatar 527

    Remove the last reference to AuctionServerManager from AuctionEntry, instead injecting a 'resolver' interface, which resolves auction ids to a server.

    by mrs, 30 Jun, 2008 07:09 AM
  • @516
    avatar 516

    Extricating UI elements into their own classes to try to reduce complexity and decrease tying.

    by mrs, 20 Jun, 2008 07:17 PM
  • @506
    avatar 506

    Split the thumbnail path-related functionality out of the thread that downloads the thumbnails, to clean up another tangle.

    by mrs, 20 Jun, 2008 12:04 AM
  • @499
    avatar 499

    Sweeping changes to look for auction entries by Identifier from the database, instead of in-memory data structures, look up deleted entries in the database (thus supporting 'don't reload deleted items' again), refactoring a few classes to reduce the amount of dependence on UI code from the Auctions code, fix some aspects of the web interface (including adding a favicon.ico, and a bug where the web interface couldn't load an item from the database).

    by mrs, 19 Jun, 2008 06:32 AM
  • @477
    avatar 477

    The build now can build a scripting-enabled and scripting-free (default) version simultaneously, and the code can recover from a lack of scripting being available. This is a somewhat temporary fix, but it may help quell the startup speed issues.

    by mrs, 01 May, 2008 03:54 AM
  • @476
    avatar 476

    One more shot at making scripting optional.

    by mrs, 30 Apr, 2008 08:58 PM
  • @475
    avatar 475

    Preliminary changes to allow scripting to be disconnected from the app, and started and used only if JRuby is available.

    by mrs, 30 Apr, 2008 08:36 PM
  • @467
    avatar 467

    Actually load auctions from the database only, and don't touch (or require) the XML file if there is some data in the database. Should speed up startup as well.

    by mrs, 27 Apr, 2008 11:13 PM
  • @450
    avatar 450

    Report in detail what is going on during app startup, via messages on the splash screen's progress bar.

    Also initialize scripting _before_ initializing the various tables, so user scripts can add columns that are saved normally in the display configuration.

    by mrs, 27 Apr, 2008 04:21 AM
  • @439
    avatar 439

    More text changes from Laurence Passmore: JBidWatcher -> JBidwatcher in all user facing strings, most notably Mac OS X application name.

    by mrs, 22 Apr, 2008 01:37 AM