On Wed, 24 Mar 2021 20:44:26 GMT, Alexey Ivanov <aiva...@openjdk.org> wrote:
>> Since this is windows specific code, I am not sure if system will not have >> any printers. I guess by default, Microsoft XPS Document Writer, Microsoft >> Print-to-PDF, Fax are present in printers list. That is the reason we do not >> have getPrintService() return 0 printer although there may not be any real >> printer present in windows system...reason for some failure in jtreg test >> which caused us to use `@key printer `tag in those tests to make real >> printers are configured. > >> Since this is windows specific code, I am not sure if system will not have >> any printers. I guess by default, Microsoft XPS Document Writer, Microsoft >> Print-to-PDF, Fax are present in printers list. > > Yes, that's correct. But you can remove them. You can also configure the > image so that these are not installed by default. > >> That is the reason we do not have getPrintService() return 0 printer >> although there may not be any real printer present in windows >> system...reason for some failure in jtreg test which caused us to use `@key >> printer `tag in those tests to make real printers are configured. > > I agree that getting zero printers in Windows is unlikely. But it's still > possible. > > Consider the following scenario: Let's assume there are 10 printers in the > system. The user removes 5 of them. In this case, `invalidateService()` is > called on the instances of `Win32PrintService` which were removed from the > system. > > Then the user removes the remaining 5 printers. In this case, > `invalidateService()` is not called at all. If an application has references > to any of these instances, they will continue to appear operational. However, > the flag `isInvalid` in `Win32PrintService` is used in two methods only: > `getPrinterState` and `getPrinterStateReasons`. > > This fix is minor, probably this situation never occurs in the real life. > > The difference in handling the deleted services caught my attention. If > everyone agrees it's not problem, I'll withdraw the PR and close the bug as > _Not an Issue_. Change looks like good practice and needed, eventhough it may occur rarely. ------------- PR: https://git.openjdk.java.net/jdk/pull/3151
