[mpeg-OTspec] Proposed update of the 'head' table flags

Terence Dowling terry at tdowling.com
Thu May 15 06:25:40 CEST 2014

I think there are two parts to this:

1) What the specification should say
2) What the accompanying information provides as guidance
     for font developers.

For the normative specification:
A) DSIG is deprecated. No font consuming software
    may place any value on the correctness of a

B) A valid font may
       1) omit DSIG entirely
        2) have a valid DSIG
        3) an empty DSIG
         4) a DSIG with a mismatched signature/content.

C) Some legacy font processors depend on the presence
     of DSIG. No new font processing software may continue this

For font developer guidance:

2) Certain older font processors use the presence of
     DSIG to trigger font treatment as "OpenType" as
    differentiated from legacy "TrueType".

     An empty DSIG will satisfy the requirements of the
    legacy processors. Developers are encouraged to
    entirely omit DSIG unless compatibility with this
     legacy software is required in which case an empty
     DSIG is sufficient and will reduce font size and font
      construction complexity. (See Note 1)

Note 1: I would hope that those responsible for the
     legacy software that uses DSIG presence would be
     forthcoming about the range (which OS, which versions
     etc.) so that appropriate information is available
     rather than a vague "maybe it is needed for something".

Note 2: Other proposals have been made for new work to reduce
     font size. This is a low cost and low risk font size reduction
     that does not require any new technology.

Note 3: Continued presence of a "Security Feature" that cannot
      be relied upon is not good practice. It provides an "appearance"
      of a protection that does not exist. If any consumer relies on this
      protection - because we did not guide them otherwise - and was
      harmed it would not reflect well on us.
       It seems well established that an important number of existing
       fonts have invalid or absent DSIG and the community seems to
       require that these fonts remain valid and must not be rejected
       because of the absence of a correct DSIG.

More information about the mpeg-otspec mailing list