----- Forwarded message from William Immendorf <[email protected]>
Date: Sat, 14 Mar 2009 21:35:41 -0500 Hey Ag, A while back, Colin Watson (Debian's Groff maintainer and the maintainer of Man-DB) and I had an email relay between each other about the update to Groff 1.20.1. Recently, Colin responded with more info about new multibyte capable Groff, and also told me about a new version of Man-DB. I know that I should put this on lfs-dev, but it got rejected for no reason. So I decided to forward it to you. I hope you find this email useful. William ---------- Forwarded message ---------- From: Colin Watson <[email protected]> Date: Mon, Feb 23, 2009 at 9:15 PM Subject: Re: Groff update, are you acepeting it? To: William Immendorf <[email protected]> On Thu, Jan 08, 2009 at 01:32:11AM +0000, William Immendorf wrote: > On 1/8/09, Colin Watson <[email protected]> wrote: > > In the past the worst problem was dealing with the differing Unicode > > support, which required serious porting of the CJK patch but which > > didn't quite offer all the facilities I needed to do a good job. > > Nowadays, preconv's ability to convert to groff Unicode entities might > > be enough to let us do a good job of this with significantly less code. > > That is a major groff improvment, I will tell LFS about that. While man-db 2.5.3 could have been persuaded to deal with it, man-db 2.5.4 (just released) adds a few tweaks to support groff 1.20 slightly more neatly. > > The remaining headache that doesn't seem to be handled by groff 1.20 is > > persuading the CJK patch to deal with kinsoku shori (Japanese > > line-wrapping semantics). I'll have a look at this and see whether I can > > bodge that in now without the rest of the seriously complicated stuff. > > In short, it's not a five-minute job, but I will definitely see what I > > can do, and am more optimistic than I was about previous versions. > > You do your best, and when you have the chance, make sure to send a > patch back, I will use that to make the LFS patch. Don't hold your breath; this is not as trivial as all that. I'm still looking at it on and off, but it may take a while. In the meantime, simply discarding the Debian multibyte patch means that CJK support *almost* works with current man-db and groff. The remaining problems are: * groff doesn't know that CJK characters are double-width, so you get the effect where each line of output is a little under twice as long as the terminal; * groff doesn't know the rules for which characters can begin or end lines (kinsoku shori, as above). * I think groff is also missing the rules for kerning between CJK and Latin characters, although I'd have to check that. All these problems ought to be solved by way of character classes, which is what I'm working on at the moment. I'll be sending updates to the groff mailing list once I have something to show. However, in the meantime, you might decide that the above problems are not critical (maybe you don't have many Japanese users, or maybe they can be tolerant of wrapping problems for a while), and in that case groff 1.20.1 should be perfectly workable for you. Regards, -- Colin Watson [[email protected]] ----- End forwarded message ----- -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
