Empty attribute value should be valid

[noqsi]

Right now, attributes with empty values are not allowed. But an empty string is a perfectly valid string, and can be useful for back ends driving downstream flows for which blank input is valid.

[vzh]

I agree with this. Besides, such attributes could work as templates, e.g. in spice backends.

[graahnul-grom]

@noqsi @vzh Although I agree with this, too, doing so right now will cause interoperability problems with geda-gaf. Schematics created in lepton
cannot be processed by that suite (gschem attributes with empty values
will be treated just like text, gnetlist will fail).
What about postponing the discussion and reopening the issue after
some major release?

[vzh]

Nothing will prevent users to create attributes in backward compatible manner.

[graahnul-grom]

Yes, but one have to always keep it in mind while working on schematics.
Allowing an empty values for attributes seems to me as something akin to file format change.