[OFFTOPIC] Info files vs Man pages

[OFFTOPIC] Info files vs Man pages

[Alex Buell]

Please, please, can we have an option to disable generation of info files
completely and an option to enable generation of man pages. 

That way, I can do --disable-info-files --enable-man-pages, and not bloat
my system with info files which I never read - I find man pages superior. 

Am I right in assuming man and info files are generated from the same
source files? If not, IHMO, this would be the ideal way to go so people
can have the option to choose to install both man/info files or to install
one of the two. 

Cheers,
Alex.

---
 /\_/\  Legalise cannabis now! 
( o.o ) Grow some cannabis today! 
 > ^ <  Peace, Love, Unity and Respect to all.

Check out http://www.tahallah.demon.co.uk

Re: [OFFTOPIC] Info files vs Man pages

[Joe Buck]

> Please, please, can we have an option to disable generation of info files
> completely and an option to enable generation of man pages. 

Well, man pages aren't generated at the moment, so you can do the former
but not the latter.

And some might find it more useful to generate HTML rather than info
files (you can generate HTML from Texinfo source with texi2html).

> That way, I can do --disable-info-files --enable-man-pages, and not bloat
> my system with info files which I never read - I find man pages superior. 

man pages typically tell you how to invoke a program, while a texinfo
document is a book, so they serve different purposes.

> Am I right in assuming man and info files are generated from the same
> source files? 

No.

Re: [OFFTOPIC] Info files vs Man pages

[Dave Love]

>>>>> "Joe" == Joe Buck <jbuck@synopsys.com> writes:

 Joe> And some might find it more useful to generate HTML rather than
 Joe> info files (you can generate HTML from Texinfo source with
 Joe> texi2html).

I never understand the advantage.  texi2html doesn't actually preserve
all the information from the Texinfo source, though it now deals
better with the linkage information.  (I can't remember if the Texinfo
4 version of makeinfo will do any better.)

Regardless, is there any HTML browser that provides the relevant
functionality (per the Info readers)?  Emacs/W3 could without great
effort, but doing `Info-search' correctly in split documents would
presumably be slow.

Re: [OFFTOPIC] Info files vs Man pages

[Alex Buell]

On 15 Sep 1998, Dave Love wrote:

> Regardless, is there any HTML browser that provides the relevant
> functionality (per the Info readers)?  Emacs/W3 could without great
> effort, but doing `Info-search' correctly in split documents would
> presumably be slow.

Lynx works a treat.

Cheers,
Alex.

---
 /\_/\  Legalise cannabis now! 
( o.o ) Grow some cannabis today! 
 > ^ <  Peace, Love, Unity and Respect to all.

Check out http://www.tahallah.demon.co.uk

Re: [OFFTOPIC] Info files vs Man pages

[Dave Love]

>>>>> "Alex" == Alex Buell <alex.buell@tahallah.demon.co.uk> writes:

 Alex> Lynx works a treat.

I use (the current version of) Lynx.  I'm pretty sure it lacks the
facilities I'm talking about.  Talk of `man -k' just indicates a lack
of understanding of Info.

Re: [OFFTOPIC] Info files vs Man pages

[Joe Buck]

> I use (the current version of) Lynx.  I'm pretty sure it lacks the
> facilities I'm talking about.  Talk of `man -k' just indicates a lack
> of understanding of Info.

Other than the ability to search across multiple nodes, what features
are you missing from Lynx when dealing with texi2html output?

Re: [OFFTOPIC] Info files vs Man pages

[Dave Love]

>>>>> "Joe" == Joe Buck <jbuck@Synopsys.COM> writes:

 Joe> Other than the ability to search across multiple nodes, what
 Joe> features are you missing from Lynx when dealing with texi2html
 Joe> output?

Apart from _regexp_ searching within a node, indexing comes to mind.
And a command set for info-like navigation.  Presumably proper
formatting of @multitable, though I'd have tried to make the output as
Lynx-friendly as possible.  Cross-references between Texinfo
documents.

I put some notes about missing features or fixmes in the texi2html
source when I'd had more than enough of Perl but I'm not sure what was
added/changed since.

[Note that rms doesn't want to use texi2html other than as a stopgap.
Apart from anything else, its copyright is unclear.]

Re: [OFFTOPIC] Info files vs Man pages

[Jeffrey A Law]

  In message < Pine.LNX.3.96.980914170353.150A-100000@lo-pc3035a >you write:
  > Am I right in assuming man and info files are generated from the same
  > source files? If not, IHMO, this would be the ideal way to go so people
  > can have the option to choose to install both man/info files or to install
  > one of the two. 
I do not think the man pages are currently generated from the info
files.  That would certainly be a better way to handle them.

I never use the info pages either, though I'm not all that hot on the
idea of yet another configure option to deal with this kind of issue.

Thoughts from the rest of this list on adding a new configure time
option to disable installation of the info files?

jeff

Re: [OFFTOPIC] Info files vs Man pages

[Alex Buell]

On Tue, 15 Sep 1998, Jeffrey A Law wrote:

> I never use the info pages either, though I'm not all that hot on the
> idea of yet another configure option to deal with this kind of issue.

If it's too much work don't bother. I normally do a rm -rf /usr/info. :o)
 
> Thoughts from the rest of this list on adding a new configure time
> option to disable installation of the info files?

I'm very sure I have seen a texi2man converter somewhere, anyone knows if
such a beast exists? Then it could well be a matter of putting in a shell
script to convert the texi files into man format & install. This, I think
is probably the easiest option. 

Also, I'm sure the man stuff could be improved by allowing hypertext links
(a la Lynx) so one can move around. Most man pages have alternative names
at the bottom of the man page, makes sense to do this :o) 

PS I just had another re-think. I think adding a configure option to
disable generation and installation of info files would be a cool idea -
the idea behind this is to save time on building egcs/glibc2. I know many
of you often do daily cvs rebuilds - disabling the texi stuff might be a
bonus as texi stuff doesn't change very much over time. 
 
Cheers, Alex.

---
 /\_/\  Legalise cannabis now! 
( o.o ) Grow some cannabis today! 
 > ^ <  Peace, Love, Unity and Respect to all.

Check out http://www.tahallah.demon.co.uk

Re: [OFFTOPIC] Info files vs Man pages

[Kamil Iskra]

On Tue, 15 Sep 1998, Alex Buell wrote:

> I'm very sure I have seen a texi2man converter somewhere, anyone knows if
> such a beast exists? Then it could well be a matter of putting in a shell
> script to convert the texi files into man format & install. This, I think
> is probably the easiest option. 

Perhaps, but what purpose would it serve? Do you really suggest that a
1.5+ MB man page is useful? It's just a waste of space on your drive and a
waste of time if you try to find something there.

man pages are supposed to provide clear, concise reference information
about commandline options and stuff like that. On the contrary, EGCS info
files contains GNU GPL, "look and feel" manifesto, description of RTL, md,
tm etc. Do you really suggest to put them in a man page?

> Also, I'm sure the man stuff could be improved by allowing hypertext links
> (a la Lynx) so one can move around.

But they wouldn't be man pages anymore. They would essentially turn into
info pages.

/ Kamil Iskra    AmigaOS  Linux/i386  Linux/m68k               \
| GeekGadgets m68k-amigaos GCC maintainer                      |
| iskra@student.uci.agh.edu.pl  kiskra@ernie.icslab.agh.edu.pl |
\ kamil@dwd.interkom.pl   http://student.uci.agh.edu.pl/~iskra /

Re: [OFFTOPIC] Info files vs Man pages

[Alex Buell]

On Wed, 16 Sep 1998, Kamil Iskra wrote:

> man pages are supposed to provide clear, concise reference information
> about commandline options and stuff like that. On the contrary, EGCS info
> files contains GNU GPL, "look and feel" manifesto, description of RTL, md,
> tm etc. Do you really suggest to put them in a man page?

I don't care. If the text console version of the info program didn't suck
so badly, I might be thinking better thoughts about info files, but I
don't, so RMS can go suck his left nut for dreaming up such a horrid user
interface. (And yes, I know there are alternatives like tkinfo but
they're x-based)

Hmm.. maybe I can throw away the info program and write a better one that
works the same way Lynx does. 

Cheers,
Alex
--
 /\_/\  Legalise cannabis now! 
( o.o ) Grow some cannabis today!
 > ^ <  Peace, Love, Unity and Respect to all.

http://www.tahallah.demon.co.uk - *new* - rewritten for text browser users!

Linux tahallah 2.1.121 #40 Thu Sep 10 07:26:39 EDT 1998 i486 unknown

Re: [OFFTOPIC] Info files vs Man pages

[Per Bothner]

> If the text console version of the info program didn't suck
> so badly, I might be thinking better thoughts about info files, but I
> don't, so RMS can go suck his left nut for dreaming up such a horrid user
> interface.

Remeber that the info interface is over 20 years old.
It was pretty good for its time.  It was still ahead of its time
when it was (re-)introduced in GNU Emacs in the 80's.

And a lot of it is a matter of taste and what you are used to.
I can't say I've ever gotten comfortable with the lynx user
interface.

How well does lynx support sequentially reading a structured document?
I'm fairly sure not as well as info does.

Still, I think lynx with some moderate extensions may make a
reasonable replacement for stand-alone info.   (One needed
extension is a way to select a info-like user interface ...)

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

Re: [OFFTOPIC] Info files vs Man pages

[Joe Buck]

> How well does lynx support sequentially reading a structured document?
> I'm fairly sure not as well as info does.

It would not be difficult to extend Lynx to handle the link
tags (so that keyboard accelerators can go to the next or previous
document).

Note that this is only the same as info for texi2html when run in
the one file per node mode, which is probably not optimal for use
over a network.  However, apparently the Mozilla guys are looking
into using the <link ... rel=next> tag for prefetching.  The idea
is that that tag would mark the page people would usually read next,
so it can be fetched while the user is reading the current page.

Re: [OFFTOPIC] Info files vs Man pages

[Dave Love]

>>>>> "Joe" == Joe Buck <jbuck@synopsys.com> writes:

 Joe> Note that this is only the same as info for texi2html when run
 Joe> in the one file per node mode, which is probably not optimal for
 Joe> use over a network.

I doubt those of us who want info features (or better) are talking
about using it over a wide area network (even though the Network was
the Computer long before it was a trademark, of course).

I can't remember whether the new makeinfo will put the tree structure
info in monolithic files, but obviously it could FWIW.  I wonder if
any browser other than W3 groks and could use REL/REV attributes of
<a>.

 Joe> However, apparently the Mozilla guys are looking into using the
 Joe> <link ... rel=next> tag for prefetching.

Blimey.  It registers at last.  But that's no help to you using the
thing to read tree-structured Texinfo output.

Re: [OFFTOPIC] Info files vs Man pages

[Joe Buck]

>  Joe> Note that this is only the same as info for texi2html when run
>  Joe> in the one file per node mode, which is probably not optimal for
>  Joe> use over a network.
> 
> I doubt those of us who want info features (or better) are talking
> about using it over a wide area network (even though the Network was
> the Computer long before it was a trademark, of course).

Any extensions to web browsers should be generally useful features.
And why *shouldn't* you be able to read info files over a wide area
network?  Why does everyone on the net have to have a local copy of
every manual?

>  Joe> However, apparently the Mozilla guys are looking into using the
>  Joe> <link ... rel=next> tag for prefetching.
> 
> Blimey.  It registers at last.  But that's no help to you using the
> thing to read tree-structured Texinfo output.

Reading Texinfo output is only one use.  Other documents prepared in other
ways also often have a tree structure.  Up to now, no one has done much
with the LINK directives because browsers haven't done anything with them.
Once the Mozilla people get everyone thinking about what "link" means,
we'll start seeing the good applications.  Anyway, if browsers are
enhanced to make reading Texinfo manuals better, it should be done in a
general manner.

Re: [OFFTOPIC] Info files vs Man pages

[Andreas Schwab]

Alex Buell <alex.buell@tahallah.demon.co.uk> writes:

|> I don't care. If the text console version of the info program didn't suck
|> so badly

Have you actually tried any recent version of the info program?

-- 
Andreas Schwab                                      "And now for something
schwab@issan.informatik.uni-dortmund.de              completely different"
schwab@gnu.org

Re: [OFFTOPIC] Info files vs Man pages

[Alex Buell]

On 17 Sep 1998, Andreas Schwab wrote:

> |> I don't care. If the text console version of the info program didn't suck
> |> so badly
> 
> Have you actually tried any recent version of the info program?

Yes, I did. It would keep doing weird things like jump to a completely
different page from the one I wanted to see no matter what I did, so gave
up and texi2ps'd them into postscript docs and read them via ghostview.

Cheers,
Alex
--
 /\_/\  Legalise cannabis now! 
( o.o ) Grow some cannabis today!
 > ^ <  Peace, Love, Unity and Respect to all.

http://www.tahallah.demon.co.uk - *new* - rewritten for text browser users!

Linux tahallah 2.1.121 #40 Thu Sep 10 07:26:39 EDT 1998 i486 unknown

Re: [OFFTOPIC] Info files vs Man pages

[Dave Love]

>>>>> "Alex" == Alex Buell <alex.buell@tahallah.demon.co.uk> writes:

 Alex> It would keep doing weird things like jump to a completely
 Alex> different page from the one I wanted to see no matter what I
 Alex> did

What was the response to your bug report?

Re: [OFFTOPIC] Info files vs Man pages

[Alex Buell]

On 19 Sep 1998, Dave Love wrote:

>  Alex> different page from the one I wanted to see no matter what I
>  Alex> did
> 
> What was the response to your bug report?

Which one? The glibc header problem or the one about info? 

If it's the info stuff, it's a user preference - mine!

I've heard nothing from Drepper about the glibc2 header problems. I'll
re-submit the bug report again to him. 

Cheers,
Alex
--
 /\_/\  Legalise cannabis now! 
( o.o ) Grow some cannabis today!
 > ^ <  Peace, Love, Unity and Respect to all.

http://www.tahallah.demon.co.uk - *new* - rewritten for text browser users!

Linux tahallah 2.1.122 #42 Thu Sep 17 08:18:35 EDT 1998 One AMD 486 DX/4 processor, 49.77 total bogomips, 32M RAM

Re: [OFFTOPIC] Info files vs Man pages

[Nicholas J. Leon]

On Wed, 16 Sep 1998, Alex Buell wrote:

 # I don't care. If the text console version of the info program didn't suck
 # so badly, I might be thinking better thoughts about info files, but I
 # don't, so RMS can go suck his left nut for dreaming up such a horrid user
 # interface. (And yes, I know there are alternatives like tkinfo but
 # they're x-based)

Now, Alex, you mustn't keep these things bottled up inside... go ahead and
speak your mind, man! Don't be afraid! 

While I agree that the current interface is lacking in many ways, I
wouldn't put it quite so harshly.

I'm currently working (and have been for a while) a perl cgi script that
does on-the-fly converstions of info files to html. It works, but needs
work :)

G'day!

--                                          n i c h o l a s  j  l e o n
  /  elegance through simplicity   / 
 /  good fortune  through truth   /          http://mrnick.binary9.net
/ not all questions have answers /          mailto:nicholas@binary9.net
 

Re: [OFFTOPIC] Info files vs Man pages

[Alex Buell]

On Fri, 25 Sep 1998, Nicholas J. Leon wrote:

> Now, Alex, you mustn't keep these things bottled up inside... go ahead and
> speak your mind, man! Don't be afraid! 

(ROFL!) What and let RMS declare a fatwa on me for heresy?! Heh. 

> I'm currently working (and have been for a while) a perl cgi script that
> does on-the-fly converstions of info files to html. It works, but needs
> work :)

NOoooooooooooooooooooooooo... NOT PERL! ;o)

Cheers,
Alex.

---
 /\_/\  Legalise cannabis now! 
( o.o ) Grow some cannabis today! 
 > ^ <  Peace, Love, Unity and Respect to all.

Check out http://www.tahallah.demon.co.uk
NEWFLASH: Reece K. Selim (Freedows) is the Shithead of the Year. Congratulations! 
Linux lo-pc3035a 2.1.122 #25 Thu Sep 17 11:06:39 EDT 1998 i586 unknown

Re: [OFFTOPIC] Info files vs Man pages

[Scott A Crosby]

> On Wed, 16 Sep 1998, Alex Buell wrote:
> 
>  # I don't care. If the text console version of the info program didn't suck
>  # so badly, I might be thinking better thoughts about info files, but I
>  # don't, so RMS can go suck his left nut for dreaming up such a horrid user
>  # interface. (And yes, I know there are alternatives like tkinfo but
>  # they're x-based)
> 

Huh? If I want to read info files from the console or a tty, I just cat
'em together and pipe to less. In x-windows, I use xemacs.

The nicest thing about info files is that they are essentially straight
text, which means they are emminently viewable in any one of the many
editors that UNIX has. HTML doesn't have this advantage, nor that nice
regex search facilities I enjoy using. Just another advantage of
non-binary human-readable files.

Scott

Re: [OFFTOPIC] Info files vs Man pages

[Dave Love]

>>>>> "Jeff" == Jeffrey A Law <law@cygnus.com> writes:

 Jeff> I do not think the man pages are currently generated from the
 Jeff> info files.  That would certainly be a better way to handle
 Jeff> them.

I concur with the comments in g77.1, but is there any facility for
formatting Texinfo in the man layout anyhow?

 Jeff> Thoughts from the rest of this list on adding a new configure
 Jeff> time option to disable installation of the info files?

I'm appalled at the idea of encouraging people to drop the online
documentation.  More than enough don't read it as it is, at least in
the g77 world.  I, for one, use it even to read stuff I wrote (a small
fraction, but representing not insignificant effort).

It's a condition of making binary distributions that they include
g77.info*, BTW -- see the documentation.  Also (currently too few) g77
error messages reference the info documentation and Compilation mode
in either the next Emacs or the one after will hyperlink them in
rms-approved fashion, kudos to Craig.  I know some people don't use
Emacs, but presumably an Infuriating Development Environment could
grok the information.

Re: [OFFTOPIC] Info files vs Man pages

[Carlo Wood]

| I do not think the man pages are currently generated from the info
| files.  That would certainly be a better way to handle them.
| 
| I never use the info pages either, though I'm not all that hot on the
| idea of yet another configure option to deal with this kind of issue.
| 
| Thoughts from the rest of this list on adding a new configure time
| option to disable installation of the info files?
| 
| jeff

info files are a pain in the ass :).  They are the "official" choice
for the format of a manual of a GNU package I understood, but its
very hard to use and I always find myself using `grep' on the raw
info files :/.

I don't think that disabling the installation of the files is the
a solution however.  Maybe a new tool to read the info files would
make it more interesting to use them.

Personally I think that a html version of docs is certainly a
*better* choice then the "info" file format.  What would be the
correct way to get html-ized docs the prefered format for GNU
packages?  Is it an option to start to include the egcs docs
as html?

-- 
 Carlo Wood  <carlo@runaway.xs4all.nl>

Re: [OFFTOPIC] Info files vs Man pages

[Jeffrey A Law]

  In message < 199809151704.TAA06119@jolan.ppro >you write:
  > info files are a pain in the ass :).  They are the "official" choice
  > for the format of a manual of a GNU package I understood, but its
  > very hard to use and I always find myself using `grep' on the raw
  > info files :/.
Similarly.  I've gotten pretty adept at reading raw info files over
the yeras.

  > Personally I think that a html version of docs is certainly a
  > *better* choice then the "info" file format.  What would be the
  > correct way to get html-ized docs the prefered format for GNU
  > packages?
You'd have to bring this up with RMS.  This isn't a decision that the
egcs project can make for the GNU world.

  > Is it an option to start to include the egcs docs as html?
We could probably make it an additional option if someone contributes
the code.  

jeff

Re: [OFFTOPIC] Info files vs Man pages

[Per Bothner]

>  > Personally I think that a html version of docs is certainly a
>  > *better* choice then the "info" file format.  What would be the
>  > correct way to get html-ized docs the prefered format for GNU
>  > packages?
> You'd have to bring this up with RMS.  This isn't a decision that the
> egcs project can make for the GNU world.

I don't think it would be difficult to get RMS and the rest of the
GNU world to accept html as a replacement for info [note I am
talking about replacing info, not texinfo], *if* if following exist:

* A program, written in C, that can translate texinfo to html,
providing most of the options and functionality of the existing makeinfo.
(texi2html is written in perl, and is therefore unsuitable for this
kind of core software.)  (I have heard rumors about people extending
makeinfo to do this, but I don't know how this is progressing.)

* A stand-alone html browser that has all or most of the
functionality of the standalone info browser (such
as searching), and that has a similar look and feel (at
least as an option).

* An emacs mode that can be used for browsing html, and
that has functionality and look and feel similar to the
existing emacs info mode.  The existing W3 mode can probably
be used as a starting-point.

It is best if all of these are or can be GPL and copyright the FSF.

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

Re: [OFFTOPIC] Info files vs Man pages

[Joe Buck]

> I don't think it would be difficult to get RMS and the rest of the
> GNU world to accept html as a replacement for info [note I am
> talking about replacing info, not texinfo], *if* if following exist:
> 
> * A program, written in C, that can translate texinfo to html,
> providing most of the options and functionality of the existing makeinfo.
> (texi2html is written in perl, and is therefore unsuitable for this
> kind of core software.)  (I have heard rumors about people extending
> makeinfo to do this, but I don't know how this is progressing.)

That's doable.

> * A stand-alone html browser that has all or most of the
> functionality of the standalone info browser (such
> as searching), and that has a similar look and feel (at
> least as an option).

Waste of time.  Extend Lynx if necessary (Lynx is under the GPL).
I can't think of anyone who prefers the standalone info browser
to Lynx that has used both.

Millions of people use HTML tools; hardly anyone uses info tools.
Given this, you don't bend the HTML tools to look like info tools;
rather you make the info data accessible to the HTML users.

> * An emacs mode that can be used for browsing html, and
> that has functionality and look and feel similar to the
> existing emacs info mode.  The existing W3 mode can probably
> be used as a starting-point.

The existing W3 mode should be sufficient (possibly with an extension).

The only extension that people have been asking for is the ability to
search across multiple files; it would be possible to come up a convention
to use META tags for this purpose (to mark the URL corresponding to
the next file).

> It is best if all of these are or can be GPL and copyright the FSF.

There's no need to rewrite Lynx just so the FSF can own it.  W3 meets
your conditions, as would an extended makeinfo.

Re: [OFFTOPIC] Info files vs Man pages

[Per Bothner]

> Waste of time.  Extend Lynx if necessary (Lynx is under the GPL).

OK, I wasn't sure about that.

> I can't think of anyone who prefers the standalone info browser
> to Lynx that has used both.

I have used both and prefer info, perhaps mainly because I use info a
lot more.

However, a few tweaks to lynx would help.  Mainly, if it could be
more intelligent about the navigation, and recognize the various
<link> tags texi2html emits, I think it would be nicer. I do
like the "intelligent space" command of info mode, and
I would like that for lynx.  Having that, plus searching,
would help convince peopel to switch.  (A traditional info-compatible
user interface *option* would also help us old-time info users.)

> Millions of people use HTML tools; hardly anyone uses info tools.
> Given this, you don't bend the HTML tools to look like info tools;
> rather you make the info data accessible to the HTML users.

I think that is a bogus point - "hardly anyone" uses lynx,
compared to the millions who point-and-click.  I think we
can tweak the lynx interface, adding the best features of info,
and come up with something quite nice.

> The only extension that people have been asking for is the ability to
> search across multiple files; it would be possible to come up a convention
> to use META tags for this purpose (to mark the URL corresponding to
> the next file).

Well, if you look at texi2html-generated html, you will find it does
emit navigation information in the links:

http://www.cygnus.com/~bothner/kawa/kawa_2.html:

<HTML>
<HEAD>
<!-- This HTML file has been created by texi2html 1.54
     from kawa.texi on 20 March 1998 -->

<TITLE>Kawa, the Java-based Scheme system - Features</TITLE>
<link href="kawa_3.html" rel=Next>
<link href="kawa_1.html" rel=Previous>
<link href="kawa_toc.html" rel=ToC>

</HEAD>
<BODY>
<p>Go to the <A HREF="kawa_1.html">first</A>, <A HREF="kawa_1.html">previous</A>
, <A HREF="kawa_3.html">next</A>, <A HREF="kawa_11.html">last</A> section, <A HR
EF="kawa_toc.html">table of contents</A>.
<P><HR><P>

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

Re: [OFFTOPIC] Info files vs Man pages

[Joe Buck]

As for Lynx vs info: yes, Lynx could be improved.  Its table layout
algorithm sucks; there's a much better one in GNU gtbl/groff.  Just
the same, Lynx has more users than info and more people would benefit
if Lynx is improved.

> > The only extension that people have been asking for is the ability to
> > search across multiple files; it would be possible to come up a convention
> > to use META tags for this purpose (to mark the URL corresponding to
> > the next file).
> 
> Well, if you look at texi2html-generated html, you will find it does
> emit navigation information in the links:

Terrific.  (I meant to say LINK tags, I didn't know they were already there).

> <TITLE>Kawa, the Java-based Scheme system - Features</TITLE>
> <link href="kawa_3.html" rel=Next>
> <link href="kawa_1.html" rel=Previous>
> <link href="kawa_toc.html" rel=ToC>

Then all you need for a "search across multiple files" command is to say
that if you get to the end and don't find the string, you fetch the link
named in a <link ... rel=Next> directive (or rel=Previous when searching
backwards).

Since this can be an expensive operation, it should be an option.  It
can either be a distinct command "Search across multiple files" or a
modifier to the existing search command.

Re: [OFFTOPIC] Info files vs Man pages

[Dave Love]

>>>>> "Per" == Per Bothner <bothner@cygnus.com> writes:

 Per> (I have heard rumors about people extending makeinfo to do this,
 Per> but I don't know how this is progressing.)

It was merged recently, but I don't think code for the additional
Texinfo markup added in the last year or so has been incoroprated.

 Per> The existing W3 mode can probably be used as a starting-point.

I don't know why only as a starting point, other than it being an
order of magnitude too slow (despite the best efforts of Bill P) and
not having all link semantics properly defined AFAIR.

Re: [OFFTOPIC] Info files vs Man pages

[Joe Buck]

>   > Personally I think that a html version of docs is certainly a
>   > *better* choice then the "info" file format.  What would be the
>   > correct way to get html-ized docs the prefered format for GNU
>   > packages?
> You'd have to bring this up with RMS.  This isn't a decision that the
> egcs project can make for the GNU world.

You can already have HTML documents.  The GNU format is Texinfo, not
info, and it can be converted into decent HTML.  HTML is a bad "source
language" (meaning, a language to write an original document in) because
too much manual labor is required (for cross-references, tables of
contents, and such), but it's a good language to generate from other
forms.

Furthermore, Texinfo has been extended (in the latest version) to add HTML
features, link links to other URLs.

>   > Is it an option to start to include the egcs docs as html?
> We could probably make it an additional option if someone contributes
> the code.  

texi2html already exists.  All that's needed is Makefile stuff to
build the HTML forms of the documents, plus conventions for defining
where the documents are stored.  One possibility is $prefix/html.

The other issue is that texi2html gives multiple choices as to how
the generated document is split into separate files: you can split
by nodes, split by chapters, or generate one big file (in all cases
the table of contents is a separate file).

Re: [OFFTOPIC] Info files vs Man pages

[Dave Love]

>>>>> "Joe" == Joe Buck <jbuck@synopsys.com> writes:

 Joe> HTML is a bad "source language"

[...]

Indeed.

 Joe> but it's a good language to generate from other forms.

That depends.  In this context a relevant consideration is whether the
cross-referencing (hypertext) model of the source matches that of HTML
(`Dexter', somewhat restricted IIRC).  In the case of Texinfo it
doesn't (see `texinfo -invisible').  This can lead to a degree of pain
and/or invalid HTML.

 Joe> texi2html already exists.

But is GNU-deprecated.  I gather the new makeinfo is nearing alpha
test.  I presume it will still handle the g77 manual, at least, but
that work was emphatically not done with the idea of dumping Info at
this juncture!

Re: [OFFTOPIC] Info files vs Man pages

[Jason Merrill]

>>>>> Dave Love <d.love@dl.ac.uk> writes:

>>>>> "Joe" == Joe Buck <jbuck@synopsys.com> writes:

 Joe> HTML is a bad "source language"

 > [...]

 > Indeed.

 Joe> but it's a good language to generate from other forms.
...
 Joe> texi2html already exists.

 > But is GNU-deprecated.  I gather the new makeinfo is nearing alpha
 > test.  I presume it will still handle the g77 manual, at least, but
 > that work was emphatically not done with the idea of dumping Info at
 > this juncture!

FWIW, Cygnus has been planning to move to DocBook/SGML as a source
language, at least internally.  I believe the LinuxDoc folks are
moving/have moved to DocBook as well.  Free SGML tools have reached
critical mass; see

  http://nis-www.lanl.gov/~rosalia/mydocs/docbook-intro.html

Jason

Re: [OFFTOPIC] Info files vs Man pages

[Dave Love]

>>>>> "JM" == Jason Merrill <jason@cygnus.com> writes:

 JM> I believe the LinuxDoc folks are moving/have moved to DocBook as
 JM> well.

Are moving AFAICT.  (I had thought that was being referred to earlier.)

 JM>   http://nis-www.lanl.gov/~rosalia/mydocs/docbook-intro.html

Unfortunately, that doesn't solve the basic problem of the strange
lack (AFAIK) of proper documentation for this documentation format --
even proprietary stuff, pending a book release.  It would be a Good
Thing if Cygnus could remedy that in the Free world.

Sorry, for squandering list resources in this thread.  I'll stop now.

Re: [OFFTOPIC] Info files vs Man pages

[Clayton Weaver]

Note:

If I am looking at the man page for some function, and this isn't quite
the one I wanted, how do I find the one I do want if I have some keyword
that I know is mentioned on the one that I want but I don't what the
actual function name is?

With info, you do

  ctrl-s keyword

and it searches the rest of the currently open subtree for you. I can't
count how many times this has been useful with glibc functions, where I
knew the general subject area but not exactly which part of it held the
tidbit of information about correct usage in some particular context.

When you know exactly which function it is and all that you need to see
are the argument list and return value, man pages are faster. When you
know what it is about, what sort of programming it is used for, but not
the exact name or which one from among several alternatives is actually
the best function in a particular code context, info files seem to me
far more convenient (once one has used info enough to get beyond
"backspace instead of up arrow for scroll up keyboard navigation
irritation syndrome").

Regards, Clayton Weaver  cgweav@eskimo.com  (Seattle)

Re: [OFFTOPIC] Info files vs Man pages

[Alex Buell]

On Tue, 15 Sep 1998, Clayton Weaver wrote:

> If I am looking at the man page for some function, and this isn't quite
> the one I wanted, how do I find the one I do want if I have some keyword
> that I know is mentioned on the one that I want but I don't what the
> actual function name is?

man apropos

QED

Cheers,
Alex.

---
 /\_/\  Legalise cannabis now! 
( o.o ) Grow some cannabis today! 
 > ^ <  Peace, Love, Unity and Respect to all.

Check out http://www.tahallah.demon.co.uk

Re: [OFFTOPIC] Info files vs Man pages

[Tim Waugh]

On Tue, 15 Sep 1998, Clayton Weaver wrote:

> If I am looking at the man page for some function, and this isn't quite
> the one I wanted, how do I find the one I do want if I have some keyword
> that I know is mentioned on the one that I want but I don't what the
> actual function name is?

Well, you could try man -k ..

Tim.
*/

Re: [OFFTOPIC] Info files vs Man pages

[Joern Rennecke]

> If I am looking at the man page for some function, and this isn't quite
> the one I wanted, how do I find the one I do want if I have some keyword
> that I know is mentioned on the one that I want but I don't what the
> actual function name is?

'man -k keyword'
gives a list of man pages that contain keyword.

Re: [OFFTOPIC] Info files vs Man pages

[Andreas Schwab]

Joern Rennecke <amylaar@cygnus.co.uk> writes:

|> > If I am looking at the man page for some function, and this isn't quite
|> > the one I wanted, how do I find the one I do want if I have some keyword
|> > that I know is mentioned on the one that I want but I don't what the
|> > actual function name is?
|> 
|> 'man -k keyword'
|> gives a list of man pages that contain keyword.

Hmm.

$ man -k SIGSEGV
SIGSEGV: nothing appropriate

-- 
Andreas Schwab                                      "And now for something
schwab@issan.informatik.uni-dortmund.de              completely different"
schwab@gnu.org

Re: [OFFTOPIC] Info files vs Man pages

[David Ross]

On Thu, 17 Sep 1998 07:26:06 -0400 Alex Buell wrote:

> 
> On 17 Sep 1998, Andreas Schwab wrote:
> 
> > |> I don't care. If the text console version of the info program didn't suck
> > |> so badly
> > 
> > Have you actually tried any recent version of the info program?
> 
> Yes, I did. It would keep doing weird things like jump to a completely
> different page from the one I wanted to see no matter what I did, so gave
> up and texi2ps'd them into postscript docs and read them via ghostview.

Yuk!

That doesn't sound like the info program, but the links in the *.info files. Of 
course that assumes incorrect but consistent results.

I use info files daily. Occassionally, I have a bad link mostly in the index 
menus. I am a GUI/mouse junkie using Emacs --with-x as an IDE.

I just has a thought (excuse me while make a record of it in my diary....back), 
have you checked your key bindings? There have been times when I have found the 
defaults changed accross releases and the documentation hadn't caught up.

Regards,



David Ross

davidr@toadtech.com
Toad Technologies

"I'll be good! I will, I will !"