On 09.11.2020 21:46, Robert Elz wrote: > Date: Mon, 9 Nov 2020 19:05:23 +0100 > From: Kamil Rytarowski <kamil%netbsd.org@localhost> > Message-ID: <04c9e1ad-df4e-1372-74d3-a17fdd5ddc09%netbsd.org@localhost> > > | I propose to remove catman(8). > > Don't. Do you use it? Do you know anybody who uses it on NetBSD-current? I don't trust that these people are tracking or using -current that used to have broken MKCATPAGES. > > | - cat pages are not generated by default since 2012 and almost nobody > | (except me?) used them in the past few years. > > That's fine, and I have no problem with the removal of MKCATMAN (or > whatever the build.sh/make variable was). That makes one less variation > that needs to be tested, and the need to remember to include the .catN > entries in the sets lists. I'd support removing the generation of > html format as well for exactly the same reasons. In both cases anyone > who finds these formats useful (including the man pages on www.netbsd.org) > can easily still generate them. > html pages are not integrated in man.conf(5) or man(1). cat pages are integrated and preferred over man pages. > I doubt that you really know (or can possibly know) who uses anything though. > Sorry, but MKCATPAGES was constantly broken AFAIR even in releases until I stepped in. > | - This tool was removed from other BSDs by default and catman is not a > | part AFAICT of any Unix specification. > > That's irrelevant. catman isn't the only application we have that > no-one else has. > It's not a selling point to any regular user, born after A.D. 2000 to optimize reading man pages. > | - Passing the documentation through mandoc(1) enables dynamic > | customization, while cat(1) cannot do much or anything as it operates on > | pre-generated .txt files. > > All true, but irrelevant. If someone needs/desires the customisation, > they they run mandoc (or even nroff, or troff) and they can do whatever > they like. People who are happy with pre-formatted man pages can use > them. What's the harm? > Reduce dead cruft. Compatibility with other systems and reduce frustration of users (I've received the complains) who are forced to hack the default install and/or tools to behave like intended. > | Personally. I recall cat-pages to be relevant on Coherent 80286 > | UNIX-like OS, operating in real mode. > > I recall them when they were first introduced - and I can assure you that > attempting to have 20 or more people all attempting to run nroff on a 780 > (simultaneously) is no fun. 20 people running cat is unnoticeable. > OK. I would be happy to know a current vax780 setup used simultaneously by 20 people on NetBSD-current who read use it for reading the man pages. > | Next, I propose to remove cat-man support from man(1) and man,conf(5). > > Please don't, what's the point? > > I really do not understand the desire to race around removing things. > That you don't see any use for them, doesn't mean that someone else > doesn't. Sorry to repeat, but as MKCATPAGES was constantly broken, unless I committed changes in the past ~5 years, nobody used that any more on anything, possibly in more than a decade. > Stuff that is broken, and can't easily be fixed, or which > needs external resources which are no longer available can be removed. cat pages cannot be fixed easily. If someone wants to pass them through mandoc and reformat, I would consider it as a waste of time. > Stuff which simply seems old, and perhaps not used as much as it once > was should just be left alone. > Being old is not the concern, but being non-trivial for reformatting. I prefer to avoid FreeBSD-style hacks that silently disable cat pages behind the scenes. Better option in 2020 is to just forget them entirely. > I suspect that there are much better things that you could be spending > time on than this kind of thing. > Introducing features that utilize mandoc, but the first step is to drop cat-pages put their feet in the door. Preformatting cat pages is a dead concept. > kre > I remind you that we already reduce unused features from our userland. For example patch(1) had removed SCCS support silently, and nobody noticed this until today. I claim that cat-pages are way less useful than SCCS-styled patches. I sense a general difference in the view point. We are apparently trading better performance on a historical computer in possibly non-existent setup anymore in two or more decades + frustrating users vs good user experience on anything modern, customizable and compatible with other OSs. But then, it's possible to keep old catman and catpages for those who want it and defer this to pkgsrc. Do I fulfill the needs of these extinct users by this move?
Attachment:
signature.asc
Description: OpenPGP digital signature