[buildcff2vf] assumes OS/2 table of default source

[frankrolf]

After merging sources with “naturally created” (through makeotf) usWinAscent values, I noticed (via FontBakery output) that the OS/2 values written into the VF just correspond to the values found in the default source.

Should this happen? Or should cff2vf perhaps have a “bounding box of all sources” calculation?

[skef]

In theory usWinAscent values should be blended like many other VF values, with the value corresponding to the default master going into the OS_2 table and the other data going into MVAR under the hcla tag. So there could be three (most likely) problems here:

1. The sources all have the same usWinAscent value so there's nothing to blend and put into the MVAR entry. (It sounds like you've confirmed this isn't what's happening.)
2. The other values are being discarded rather than blended into an MVAR entry.
3. FontBakery is just showing the value from the table and not taking the MVAR entry into account.

[frankrolf]

This is what the VF in question has in its MVAR table, for the hcla tag:

    <ValueRecord index="0">
      <ValueTag value="hcla"/>
      <VarIdx value="1"/>
    </ValueRecord>

In other words – not much?