diff mbox

Update to WWW Docs for Contributing Doc Patches [wwwdocs]

Message ID alpine.LNX.2.00.1009241941470.5068@acrux.dbai.tuwien.ac.at
State New
Headers show

Commit Message

Gerald Pfeifer Sept. 24, 2010, 9:24 p.m. UTC
On Wed, 8 Sep 2010, Tom Browder wrote:
> When  trying to do "make info" and "make dvi" according to wwwdocs for
> doc patches, I got errors due to the build environment after a
> configure (outside the gcc tree).  I found I had to also complete a
> build of the c language to get those targets to work--thus the
> recommended change in the patch at the end of this message.
> 
> I used this minimal set of commands to get a good build of the
> required doc test targets:
> 
> $ ../<gcc trunk src>/configure --enable-languages=c --disable-bootstrap
> $ make
> $ make info
> $ make dvi

Thanks, Tom!  I suggest the small variation of your patch that I am 
including below.  Joseph, thoughts?

Gerald

Comments

Joseph Myers Sept. 24, 2010, 10:34 p.m. UTC | #1
On Fri, 24 Sep 2010, Gerald Pfeifer wrote:

> Thanks, Tom!  I suggest the small variation of your patch that I am 
> including below.  Joseph, thoughts?

This is probably OK as a simplification.  (You shouldn't need a bootstrap; 
once the gcc/ directory has been fully configured you should be able to 
run "make info" etc. in that directory in your build directory without the 
rest of the build being needed.  Maybe info-gcc and dvi-gcc should be 
recommended at top level if they work, to avoid any problems with 
recursing into other directories without a full build.)
Ralf Wildenhues Sept. 26, 2010, 6:03 a.m. UTC | #2
* Joseph S. Myers wrote on Sat, Sep 25, 2010 at 12:34:17AM CEST:
> (You shouldn't need a bootstrap; 
> once the gcc/ directory has been fully configured you should be able to 
> run "make info" etc. in that directory in your build directory without the 
> rest of the build being needed.  Maybe info-gcc and dvi-gcc should be 
> recommended at top level if they work, to avoid any problems with 
> recursing into other directories without a full build.)

This doesn't currently work.  I opened PR bootstrap/45796 for this.

Cheers,
Ralf
Gerald Pfeifer Sept. 29, 2010, 2:41 a.m. UTC | #3
On Sun, 26 Sep 2010, Ralf Wildenhues wrote:
>> (You shouldn't need a bootstrap; once the gcc/ directory has been fully 
>> configured you should be able to run "make info" etc. in that directory 
>> in your build directory without the rest of the build being needed.  
>> Maybe info-gcc and dvi-gcc should be recommended at top level if they 
>> work, to avoid any problems with recursing into other directories 
>> without a full build.)
> This doesn't currently work.  I opened PR bootstrap/45796 for this.

So, should we hold off Tom's documentation patch for now?

Gerald
Ralf Wildenhues Sept. 29, 2010, 4:41 a.m. UTC | #4
* Gerald Pfeifer wrote on Wed, Sep 29, 2010 at 04:41:24AM CEST:
> On Sun, 26 Sep 2010, Ralf Wildenhues wrote:
> >> (You shouldn't need a bootstrap; once the gcc/ directory has been fully 
> >> configured you should be able to run "make info" etc. in that directory 
> >> in your build directory without the rest of the build being needed.  
> >> Maybe info-gcc and dvi-gcc should be recommended at top level if they 
> >> work, to avoid any problems with recursing into other directories 
> >> without a full build.)
> > This doesn't currently work.  I opened PR bootstrap/45796 for this.
> 
> So, should we hold off Tom's documentation patch for now?

IMHO you shouldn't.  The text can still be simplified when the
requirement for a previous bootstrap is not needed any more for the
build environment to get set up on any of the active branches.

Thanks,
Ralf
Gerald Pfeifer Jan. 28, 2011, 6:12 p.m. UTC | #5
On Wed, 29 Sep 2010, Ralf Wildenhues wrote:
>> So, should we hold off Tom's documentation patch for now?
> IMHO you shouldn't.  The text can still be simplified when the
> requirement for a previous bootstrap is not needed any more for the
> build environment to get set up on any of the active branches.

Okay, so I went ahead and committed my proposed patch which is
based on Tom's suggestion; included below again for reference.

Thanks Tom, and sorry for the delay!

Gerald

PS: And, please, do complain if you don't see traction with any of your 
submissions; sometimes things simply may fall through the cracks.
Ralf Wildenhues Jan. 28, 2011, 7:06 p.m. UTC | #6
* Gerald Pfeifer wrote on Fri, Jan 28, 2011 at 07:12:06PM CET:
> On Wed, 29 Sep 2010, Ralf Wildenhues wrote:
> >> So, should we hold off Tom's documentation patch for now?
> > IMHO you shouldn't.  The text can still be simplified when the
> > requirement for a previous bootstrap is not needed any more for the
> > build environment to get set up on any of the active branches.
> 
> Okay, so I went ahead and committed my proposed patch which is
> based on Tom's suggestion; included below again for reference.

FWIW, I don't see the patch, but this was about
http://gcc.gnu.org/ml/gcc-patches/2010-09/msg02256.html and
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45796
which has long been fixed since.

Cheers,
Ralf
diff mbox

Patch

Index: contribute.html
===================================================================
RCS file: /cvs/gcc/wwwdocs/htdocs/contribute.html,v
retrieving revision 1.77
diff -u -3 -p -r1.77 contribute.html
--- contribute.html	10 Jul 2010 20:22:25 -0000	1.77
+++ contribute.html	24 Sep 2010 17:42:49 -0000
@@ -148,8 +148,9 @@  testsuite run.</p>
 
 <h2><a name="docchanges">Documentation Changes</a></h2>
 
-<p>Documentation changes do not require a bootstrap, but you must
-perform <code>make info</code> and <code>make dvi</code> and correct
+<p>Documentation changes do not require a new bootstrap (a working
+bootstrap is necessary to get the build environment correct), but you
+must perform <code>make info</code> and <code>make dvi</code> and correct
 any errors.  You should investigate complaints about overfull or
 underfull hboxes from <code>make dvi</code>, as these can be the only
 indication of serious markup problems, but do not feel obliged to