Follow-up Comment #5, bug #67817 (group groff): [comment #4 comment #4:] > we can't fix this regression without causing another one. > > See bug #55799.
I was wondering if that might have been a contributor.
> Approximate output can either traverse into a "composite node"
> (read: user-defined character) or it cannot.
Unilaterally? Surely (I say, not having any idea what the algorithm looks
like) it can do this conditionally. That is, try the traversal, and if it
finds itself with nothing better to output than "<--->," go back to the old
output.
Maybe this is too hard to do in a non-kludgey manner to justify the effort, in
which case, rejecting it is valid. If groff had always printed "<--->" in
these places, I doubt I'd be arguing to change it. But it used to output
something a lot more useful, and if there's a way to restore that without
major code contortions, that serves the user better.
[comment #3 comment #3:]
> No, but they *could*, and reasonably so.
I suppose I'm spending enough time responding to actual arguments that I don't
feel compelled to respond to hypothetical ones as well.
> Should we be unique over the naming scheme of *roff special
> characters or unique over their output representations?
Who cares? It's approximate. It doesn't have to be 100% unambiguous. The
point is that "<--->" is far more ambiguous than "<S trademark>". You can
reduce ambiguity by quite a bit without making guarantees of absolute
uniqueness. We've both acknowledged that -a makes few guarantees anyway.
> But the character name as defined by the document is not what
> the output driver uses to produce the glyph.
Nor is "<--->". But one of them is far more likely to be meaningful to the
reader.
> another user could, with every bit as much justice, assert
No one seems to have much opinion at all in the email thread. Maybe all the
opinionated folk besides Deri are on holiday.
> "-a" output cannot express everything and for several kinds of
> formatting object or command, it makes no attempt. Font style
> and type size changes are further examples,
Perfect is the enemy of good. You don't back away from things you _can_
represent just because there remain things you _can't_.
_______________________________________________________
Reply to this item at:
<https://savannah.gnu.org/bugs/?67817>
_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/
signature.asc
Description: PGP signature
