[squeak-dev] Should InvalidDirectoryError be a Notification?

Tim Johnson digit at sonic.net
Fri May 20 15:59:41 UTC 2022


Hi Christoph,

> On May 20, 2022, at 5:00 AM, christoph.thiede at student.hpi.uni-potsdam.de wrote:
> 
> With InvalidDirectoryError, we have one more exception class that overrides #defaultAction to handle itself silently without bringing up a debugger. As a consequence, some tests that reproduce normal user interactions will fail for no visible reason. See DirectoryChooserDialogTest>>testChooseAbsentDirectory in polish-file-dialogs.3.cs <https://lists.squeakfoundation.org/pipermail/squeak-dev/2022-May/220634.html> for an example.
> 
> I am unsure how to deal with these situations: Isn't an error that doesn't bring up a debugger some kind of anti-pattern?

I usually think of these as decisions that were made to support end-user experience in Squeak.  In this case, perhaps it supports hard-coded paths that live within older release images and/or VM builds on various platforms.  Or maybe it made loading image segments smoother...

> Should we rethink the definition of this exception class?

If it's now possible to give a test designer the choice of whether to handle the exception or not, that sounds like a win...


Best,
Tim



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20220520/247b8ba9/attachment-0001.html>


More information about the Squeak-dev mailing list