fontBuilder: Don't add a stub signature to DSIG #1621
Conversation
|
Why do we need two different fake/dummy DSIG tables? If the empty one is the more canonical way to do it, I'd be happy to just have that. |
|
I wonder if it’s actually still the case that “some MS applications require this”, do you have concrete examples with recent versions? |
|
Plenty of people are still using versions of Microsoft Office applications that work fine for them and that require the DSIG for OT features to work. |
|
Is there more information about what OT features this issue influences for MS Word users at any affected version? |
|
If we have to add one I prefer it is an empty one rather than an invalid but non-empty DSIG. And possibly disabled by default so if one cares and knows what they are doing the can opt in |
Is the public/private key authentication "feature" of applications that use this table removed and, for some reason, they simply need the table to be present to support some OT features as Denis suggested above? If DSIG certificate authentication is still occurring out there, then the approach in your comment seems mandatory Cosimo. |
An empty DSIG is enough to make MS applications happy and also fixes an error reported by MS Font Validator
3ff5217 to
9fb0cac
Compare
|
@justvanrossum an empty DSIG is enough. That's what makeotf has been doing for a while and we know of no side effects. Also, MS Font Validator complains about the stub signature. I've updated the PR. @anthrotype FontBuilder only adds a DSIG when |
Then let's make it do that instead of the current behavior. Yes, FontBuilder never adds a dummy DSIG implicitly. |
An empty DSIG is enough to make MS applications happy and also fixes an error reported by MS Font Validator.