Today I reported a bug https://quality.embarcadero.com/browse/RSP-10066 only to find out it has been already reported and allegedly fixed twice http://qc.embarcadero.com/wc/qcmain.aspx?d=111029 and http://qc.embarcadero.com/wc/qcmain.aspx?d=104035

Today I reported a bug https://quality.embarcadero.com/browse/RSP-10066 only to find out it has been already reported and allegedly fixed twice http://qc.embarcadero.com/wc/qcmain.aspx?d=111029 and http://qc.embarcadero.com/wc/qcmain.aspx?d=104035

What is going on there?
https://quality.embarcadero.com/browse/RSP-10066

Comments

  1. I downloaded your sample, installed the package and ran the demo and it is styled correctly I believe.

    ReplyDelete
  2. I attached a screenshot to your bug. Is that not how you expect it to work?

    ReplyDelete
  3. Jim McKeeth Nope. I thought that attached project would show issue correctly. 

    There is randomness in whether issue is manifested or not. If exact class cannot be found, appropriate StyleHook will be searched among ancestor classes, but code will return ANY ancestor - first one it runs into and that one may be wrong.

    Since RgisteredStyleHooks is dictionary where TClass is key, order of classes will depend on TClass hash that is basically pointer to TClass info and can change randomly as you add or remove code. And apparently even same code run on different computers can work properly or not.

    You can find more at http://stackoverflow.com/questions/28463556/vcl-styles-breaks-randomly

    And I will attach "broken" screenshot.

    ReplyDelete
  4. Thanks, I have a better idea of what is going on with the randomness.

    ReplyDelete

Post a Comment