Re: [PATCH] gdb: fix handling of raw ANSI escape sequences printed from Python

[Tom Tromey <tom@tromey.com>]

>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:

>> If the pager is already parsing escape sequences and reacting
>> accordingly, then I wonder if there's any need at all for
>> ui_file::emit_style_escape.

Andrew> I'm not entirely sure what you mean here, but I do wonder if the
Andrew> m_applied_style should be moved from ui_file, and ui_file should
Andrew> become a pure interface class.

This is done in the redirection series :)

Anyway what I meant is that I think emit_style_escape is really only
implemented in a special way by the pager.  The reason for its
existence, IIRC, was to track the styles when wrapping.

However since the pager already tracks styles in a different way, then
it seems like there's no reason for any special code.

Andrew> It's kind-of weird that the pager class ends up tracking some state
Andrew> within its grandparent's ui_file::m_applied_style, but then it also
Andrew> needs to track and mirror some state from within
Andrew> m_stream::m_applied_state.

Andrew> I wonder if there should be a new class which inherits from ui_file
Andrew> which actually implements styling, and would be used for the basic
Andrew> streams.

I think only the pager really cares so, IMO anyway, it makes sense to
leave the logic there.

FWIW I have another series in progress that adds styling to error() and
it also required some changes to ui-file, though that's mostly
templatizing some classes so they can be used in other contexts.

Andrew> I'm going to check this in to both master, and the 17 branch.

I meant to mention a 17 backport, thanks for doing that.

Tom