>>>>> "jpw" == John Paul Wallington <jpw(a)shootybangbang.com&gt; writes: jpw> Adrian Aichner <adrian(a)xemacs.org&gt; wrote: >>>>>>> "jpw" == John Paul Wallington <jpw(a)shootybangbang.com&gt; writes: >> jpw> Adrian Aichner <adrian.aichner(a)teradyne.com&gt; wrote: >> >> ... on the assumption this is an oversight. >> jpw> It was deliberate. In GNU 21, the header-line is like a modeline, jpw> only at the top of the buffer. Setting the buffer-local variable jpw> header-line-format makes it magically appear. It's background and jpw> forground colours are distinct (eg: light grey, dark grey). So having jpw> it default to font-lock-type-face doesn't look right; perhaps there is jpw> a more appropriate default? >> >> So you're basically OK with migrating it to defcustom as well? jpw> Sure. But I think it should have a specific/distinct default too. >> Since <quote author="jpw">In GNU 21, the header-line is like a >> modeline</quote>, how about 'modeline as a default value? jpw> I don't think that's a good value, because when there are >1 jpw> windows in a frame, the header-line and mode-line would appear to jpw> "merge" into a really fat modeline! I wouldn't mind but I see your point. jpw> Do you like this: jpw> Make a face called header-line (if it doesn't exist) with the present jpw> defaults, and use 'header-line as the default value within jpw> ibuffer-header-line-face's defcustom? How about this question then: Should XEmacs port the magical header-line feature of GNU Emacs 21.1?

Another thing: ibuffer (as long as it stays a packge) must work with XEmacs 21.1 as well, which will definitely not get this 'header-line feature.

Where should the 'header-line face be defined?