Re: Save and restore setup.exe window geometry attempt #2

[Dave Korn <dave dot korn dot cygwin at googlemail dot com>]

Jonathon Merz wrote:
> Hi all,
> 
> I've reworked the patch based on the advice I recieved from my last
> submission, 

  Thanks for persisting.

> My questions are:
> 1. In my current version, the GeometrySetting class just writes/reads
> a raw WINDOWPLACEMENT struct to/from the geometry settings file.  This
> keeps the code nice and succinct, but is a departure from the way the
> other settings files are handled in that the resulting file is not
> human readable/editable.  The question is: is it preferred to have the
> settings files be human readable at the expense of larger code (which
> would be entirely contained within the new GeometrySetting class's
> save() and load() functions) or is it acceptable as attached?

  I'm afraid not, but it's a simple thing to fix.  We do want the setting
files to be plain ascii text.  Makes it a lot easier to debug people's setup
problems sometimes if we can get them to post the contents.  Should be as easy
as reading the io_stream's contents into some sort of strsteambuf (but don't
quote me on that until I've had time to look up an STL reference) and using
the << or >> operators.

> 2. At this time, I've got the call to getWindowPlacement within the
> handling of the WM_SIZE message in the block in the same block as all
> of the existing resizing code, where it will deliberately not catch
> when the window is minimized, so setup will never restore in a
> minimized state.  As far as I can tell, there is not any typical way
> to exit while minimized.  You _can_ minimize the window then
> right-click and close it, but if you want to completely finish setup,
> you have to click the finish button to execute your choices on the
> last dialog.  My original choice to arrange the code this way was
> fairly arbitrary, so if anyone thinks it would be useful in some way
> to catch minimization as well, or if you just think it should catch
> minimization, by all means let me know.

  I think that not restoring minimisation is absolutely the correct thing to do.

> 3. And of course, let me know if you see any other problems, since I
> think the code is pretty close to complete pending any changes
> resulting from the answers to 1 and 2.

  I'll apply your patch to my local tree and have a play with it.

    cheers,
      DaveK