egrep -i haifa INSTALL install.texi

egrep -i haifa INSTALL install.texi

[Gerald Pfeifer]

...in egcs-971127/gcc only produces matches in install.texi, not a single
one in INSTALL. 

Shouldn't these two files be in sync? (Respectively, how about making this
automatic?)

Hope this helps,
Gerald
-- 
Gerald Pfeifer (Jerry)      Vienna University of Technology
pfeifer@dbai.tuwien.ac.at   http://www.dbai.tuwien.ac.at/~pfeifer/

Re: egrep -i haifa INSTALL install.texi

[Jeffrey A Law]

  In message < Pine.GSO.3.96.971124101147.2148B-100000@alcyone.dbai.tuwien.ac.at
>you write:
  > ...in egcs-971127/gcc only produces matches in install.texi, not a single
  > one in INSTALL. 
  > 
  > Shouldn't these two files be in sync? (Respectively, how about making this
  > automatic?)
gcc/INSTALL is basically out of date and depreciated for egcs.

If you want a peek at what the installation instructions are likely
to look like check out
http://www.cygnus.com/egcs/install/index.html

jeff

Re: egrep -i haifa INSTALL install.texi

[Joe Buck]

> gcc/INSTALL is basically out of date and depreciated for egcs.

OK.  I hope you've planned one snapshot for a debugging pass on the
included documentation (that is, it's a bad idea to do a release with
documentation that people haven't yet seen).

> http://www.cygnus.com/egcs/install/index.html

On the directions:

configure.html:

I've always built egcs with srcdir = objdir and have not had problems.
Is that warning appropriate?  objdir is certainly appropriate for the
system administrator who supports multiple Unix flavors, but why do
suspect that srcdir == objdir is risky.

I recommend that all defaults for options switches be listed;
--with-local-prefix, --with-gxx-include-dir are missing theirs.

--with-newlib is not documented; a sentence saying what newlib is would
be helpful.

build.html:

Please document bootstrap-lean:

If you are short on disk space, you might consider "make bootstrap-lean"
instead.  This is identical to "make bootstrap" except that object files
from the stage 1 and stage 2 of the 3-stage bootstrap of the compiler
are deleted as soon as they are no longer needed.

The "make cross" directions are insufficient: the user also must obtain
target tools and libraries -- a pointer to the crossgcc FAQ might be
helpful here, even if it is somewhat out of date.

test.html:

I suggest putting in the full path to the dejagnu testsuite as visible
text, not just a link, so people printing out the directions have it.


finalinstall.html:

... where <I>prefix</I> is the value you specified with the --prefix option to
configure (or /usr/local by default).

Re: egrep -i haifa INSTALL install.texi

[Jeffrey A Law]

  In message < 199712011803.KAA26566@atrus.synopsys.com >you write:
  > 
  > > gcc/INSTALL is basically out of date and depreciated for egcs.
  > 
  > OK.  I hope you've planned one snapshot for a debugging pass on the
  > included documentation (that is, it's a bad idea to do a release with
  > documentation that people haven't yet seen).
Yup, it should start spinning in the next hour or so as I go back and
start tweaking the docs included in the release.

It'll also include a small number of critical fixes.


  > I've always built egcs with srcdir = objdir and have not had problems.
  > Is that warning appropriate?  objdir is certainly appropriate for the
  > system administrator who supports multiple Unix flavors, but why do
  > suspect that srcdir == objdir is risky.
We've had trouble with it in the past -- multilibs haven't played well when
srcdir == objdir for example.


  > I recommend that all defaults for options switches be listed;
  > --with-local-prefix, --with-gxx-include-dir are missing theirs.
Thanks.  Done.

  > --with-newlib is not documented; a sentence saying what newlib is would
  > be helpful.
Done.

  > Please document bootstrap-lean:
Thanks.  I used your text.

  > The "make cross" directions are insufficient: the user also must obtain
  > target tools and libraries -- a pointer to the crossgcc FAQ might be
  > helpful here, even if it is somewhat out of date.
Hmmm, I thought I had the link.  I'll go figure out what happened :-)
I suspect they will stay insufficient for a while until we get a better
handle on how to document all the pieces that folks need to build useable
cross tools.

  > test.html:
  > 
  > I suggest putting in the full path to the dejagnu testsuite as visible
  > text, not just a link, so people printing out the directions have it.
Sounds reasonable.  Done.

  > finalinstall.html:
  > 
  > ... where <I>prefix</I> is the value you specified with the --prefix option
  >  to
  > configure (or /usr/local by default).
Thanks.

jeff

Re: egrep -i haifa INSTALL install.texi

[Gerald Pfeifer]

On Sun, 30 Nov 1997, Jeffrey A Law wrote:
>> ...in egcs-971127/gcc only produces matches in install.texi, not a single
>> one in INSTALL. 
> gcc/INSTALL is basically out of date and depreciated for egcs.

Well, but in that case it should be either completely removed or --
probably better -- contain a pointer to install.texi:

  "This file, which used to contain installation instructions for
   gcc/egcs, has been superseded by install.texi. Please check the
   information there."


IMHO, out of all possible scenarios, an out-of-date gcc/INSTALL is surely
the worst: Once having found INSTALL, most people won't look any further,
thus using obsolete resp. incomplete information.

Gerald
-- 
Gerald Pfeifer (Jerry)      Vienna University of Technology
pfeifer@dbai.tuwien.ac.at   http://www.dbai.tuwien.ac.at/~pfeifer/

Re: egrep -i haifa INSTALL install.texi

[Jeffrey A Law]

  In message <Pine.GSO.3.96.971202223251.14537A-100000@alcyone.dbai.tuwien.ac.a
t>you write:
  > Well, but in that case it should be either completely removed or --
  > probably better -- contain a pointer to install.texi:
No, because there is still some useful information in there -- like installing
on VMS for example, or other (mostly old) oddball machines that couldn't
bootstrap gcc directly.


  > IMHO, out of all possible scenarios, an out-of-date gcc/INSTALL is surely
  > the worst: Once having found INSTALL, most people won't look any further,
  > thus using obsolete resp. incomplete information.
They find INSTALL in the toplevel first.  For 99% of the folks using egcs
it should be enough.


jeff

Re: egrep -i haifa INSTALL install.texi

[H.J. Lu]

> 
> On Sun, 30 Nov 1997, Jeffrey A Law wrote:
> >> ...in egcs-971127/gcc only produces matches in install.texi, not a single
> >> one in INSTALL. 
> > gcc/INSTALL is basically out of date and depreciated for egcs.
> 
> Well, but in that case it should be either completely removed or --
> probably better -- contain a pointer to install.texi:
> 
>   "This file, which used to contain installation instructions for
>    gcc/egcs, has been superseded by install.texi. Please check the
>    information there."
> 
> 
> IMHO, out of all possible scenarios, an out-of-date gcc/INSTALL is surely
> the worst: Once having found INSTALL, most people won't look any further,
> thus using obsolete resp. incomplete information.
> 

Here is the patch in gcc2. It is very straight forward to rebuild
INSTALL from install.texi. Why not do it?


H.J.
---
Sun Aug  3 17:27:44 1997  H.J. Lu  (hjl@gnu.ai.mit.edu)

	* Makefile.in (INSTALL): Build in $(srcdir).

--- Makefile.in.orig	Tue Dec  2 18:43:46 1997
+++ Makefile.in	Tue Dec  2 18:44:17 1997
@@ -2011,8 +2011,8 @@
 
 
 INSTALL: $(srcdir)/install1.texi $(srcdir)/install.texi
-	$(MAKEINFO) -D INSTALLONLY --no-header --no-split \
-	$(srcdir)/install1.texi -o INSTALL
+	cd $(srcdir); $(MAKEINFO) -D INSTALLONLY --no-header \
+		--no-split install1.texi -o INSTALL
 #\f
 # Deletion of files made during compilation.
 # There are four levels of this: