On Thu, 7 Mar 2024 00:47:00 GMT, Phil Race <p...@openjdk.org> wrote: >> It is not what I meant… If we want to preserve the way it work, we should >> execute the below code even if `this->GetDC()` fails… >> >> On the other hand, testing didn't find any problems, in which case I prefer >> bailing out after `::GetDIBits` fails too. >> >> @prrace, Do you know how critical the code below is? Is it worth to code the >> fallback if the path to `::GetDIBits`? > >> @prrace, Do you know how critical the code below is? Is it worth to code the >> fallback if the path to ::GetDIBits? > > In the headLESS case we should never be here, since this is initializing the > devices that don't exist in that case. > In the headFUL case I would prefer to keep the existing non-debug behaviour > that we soldier on and try to do something sensible. Line 204 has a comment > about a case in which GetDIBits really did fail and there's code to cope. I > am writing specifically about GetDIBits.
OK, I have reverted all changes to awt_Win32GraphicsDevice.cpp since we don't go through the code in the new headless case any more. I guess it's best to leave any refactoring of this code to later efforts. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/17614#discussion_r1516492296
