[RFA] gdb_ari.sh changes to discard some file types for specific tests

[RFA] gdb_ari.sh changes to discard some file types for specific tests

[Pierre Muller]

  UCASE and EditCase tests are currently polluted
by lex and yacc files.
  This patch discards those bugs if the source 
is a lex or yacc source.
  The patch also contains the fixes for xasprintf and xvasprintf
that I already proposed yesterday.

  I am new to AWK scripts,
so I am unsure if my changes are 
standard.

Is this OK?

Pierre Muller

Pseudo ChangeLog:

2007-10-16  Pierre Muller  <muller@ics.u-strasbg.fr>
        * gdb_ari.sh (is_yacc_or_lex): New variable.
	  (FNR == 1): Set is_yacc_or_lex according to FILENAME
	  pattern matching;
	  (UCASE): Discard error if is_yacc_or_lex is set.
	  (editCase): Ditto. Give more precise documentation.
	  (xasprintf): Add known entries in defs.h and utils.c.
  	  Switch to regression category.
        (xvasprintf): Idem.

$ cvs diff -up gdb_ari.sh
Index: gdb_ari.sh
===================================================================
RCS file: /cvs/gdbadmin/ss/gdb_ari.sh,v
retrieving revision 1.75
diff -u -p -r1.75 gdb_ari.sh
--- gdb_ari.sh  10 Oct 2007 11:56:14 -0000      1.75
+++ gdb_ari.sh  16 Oct 2007 12:34:18 -0000
@@ -172,6 +172,13 @@ function fail(bug,supplement) {

 FNR == 1 {
     seen[FILENAME] = 1
+    if (match(FILENAME, "\\.[ly]$") || match(FILENAME, "-lex\\.c")) {
+      # FILENAME  is  a lex or yacc source, or a C file generated by lex
+      is_yacc_or_lex = 1
+    }
+    else {
+      is_yacc_or_lex = 0
+    }
 }
 END {
     if (print_idx) {
@@ -455,15 +462,19 @@ Function name is uppercase."
     category["UCASE function"] = ari_code
 }
 /^[A-Z][[:alnum:]_]*[[:space:]]*\(/ {
-    fail("UCASE function")
+   if (is_yacc_or_lex == 0) {
+      fail("UCASE function")
+    }
 }

 BEGIN { doc["editCase function"] = "\
-Function name is uppercase."
+Function name starts lower case but has uppercased letters."
     category["editCase function"] = ari_code
 }
-/^[a-z][a-z0-0_]*[A-Z][a-z0-9A-Z_]*[[:space:]]*\(/ {
-    fail("editCase function")
+/^[a-z][a-z0-9_]*[A-Z][a-z0-9A-Z_]*[[:space:]]*\(/ {
+   if (is_yacc_or_lex == 0) {
+     fail("editCase function")
+   }
 }

 BEGIN { doc["hash"] = "\
@@ -1300,7 +1311,9 @@ Do not use vasprintf(), instead use xstr

 BEGIN { doc["xasprintf"] = "\
 Do not use xasprintf(), instead use xstrprintf"
-    category["xasprintf"] = ari_code
+    fix("xasprintf", "gdb/defs.h", 1)
+    fix("xasprintf", "gdb/utils.c", 1)
+    category["xasprintf"] = ari_regression
 }
 /(^|[^_[:alnum:]])xasprintf[[:space:]]*\(/ {
     fail("xasprintf")
@@ -1308,7 +1321,9 @@ Do not use xasprintf(), instead use xstr

 BEGIN { doc["xvasprintf"] = "\
 Do not use xvasprintf(), instead use xstrvprintf"
-    category["xvasprintf"] = ari_code
+    fix("xvasprintf", "gdb/defs.h", 1)
+    fix("xvasprintf", "gdb/utils.c", 1)
+    category["xvasprintf"] = ari_regression
 }
 /(^|[^_[:alnum:]])xvasprintf[[:space:]]*\(/ {
     fail("xvasprintf")

Re: [RFA] gdb_ari.sh changes to discard some file types for specific tests

['Daniel Jacobowitz']

On Tue, Oct 16, 2007 at 02:41:34PM +0200, Pierre Muller wrote:
>   UCASE and EditCase tests are currently polluted
> by lex and yacc files.
>   This patch discards those bugs if the source 
> is a lex or yacc source.
>   The patch also contains the fixes for xasprintf and xvasprintf
> that I already proposed yesterday.
> 
>   I am new to AWK scripts,
> so I am unsure if my changes are 
> standard.
> 
> Is this OK?

I checked in something similar - we shouldn't be looking at lex/yacc
output files at all.

-- 
Daniel Jacobowitz
CodeSourcery

RE: [RFA] gdb_ari.sh changes to discard some file types for specific tests

[Pierre Muller]

Thanks!

  I have more changes, especially
related to doscheck.c
(which is not used on the web pages,
but probably should be enabled again.)
  I changed the output of doscheck to
avoid getting false duplicates for files
in different directories.

  The next question is for Eli, as DJGPP maintainer:
  I have locally a working version of doscheck,
which avoid duplicates in different directories,
but still reveals some conflicts for DJGPP.
  The most notable is
gdb/ChangeLog and gdb/ChangeLog-2006
but there are others in bfd, libdecnumber, gdb/testsuite
and other directories,
  but lots of these duplicates are probably
in directories that are not even put in
the DJGPP gdb sources.

  Should I send the whole list,
or should we start by updating the
directory exclusion list in update_web_ari?

  By the way, is the script that
generates the DJGPP gdb source public?
If yes, where can I find it?

Pierre

> -----Original Message-----
> From: gdb-patches-owner@sourceware.org [mailto:gdb-patches-
> owner@sourceware.org] On Behalf Of 'Daniel Jacobowitz'
> Sent: Wednesday, October 24, 2007 10:25 PM
> To: Pierre Muller
> Cc: gdb-patches@sourceware.org
> Subject: Re: [RFA] gdb_ari.sh changes to discard some file types for
> specific tests
> 
> On Tue, Oct 16, 2007 at 02:41:34PM +0200, Pierre Muller wrote:
> >   UCASE and EditCase tests are currently polluted
> > by lex and yacc files.
> >   This patch discards those bugs if the source
> > is a lex or yacc source.
> >   The patch also contains the fixes for xasprintf and xvasprintf
> > that I already proposed yesterday.
> >
> >   I am new to AWK scripts,
> > so I am unsure if my changes are
> > standard.
> >
> > Is this OK?
> 
> I checked in something similar - we shouldn't be looking at lex/yacc
> output files at all.
> 
> --
> Daniel Jacobowitz
> CodeSourcery

Re: [RFA] gdb_ari.sh changes to discard some file types for specific tests

['Daniel Jacobowitz']

On Thu, Oct 25, 2007 at 09:17:54AM +0200, Pierre Muller wrote:
>   The next question is for Eli, as DJGPP maintainer:
>   I have locally a working version of doscheck,
> which avoid duplicates in different directories,
> but still reveals some conflicts for DJGPP.
>   The most notable is
> gdb/ChangeLog and gdb/ChangeLog-2006
> but there are others in bfd, libdecnumber, gdb/testsuite
> and other directories,

This means you should probably update config/djgpp/fnchange.lst.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [RFA] gdb_ari.sh changes to discard some file types for specific tests

[Eli Zaretskii]

> Date: Thu, 25 Oct 2007 08:09:29 -0400
> From: 'Daniel Jacobowitz' <drow@false.org>
> Cc: 'Eli Zaretskii' <eliz@gnu.org>, gdb-patches@sourceware.org
> 
> On Thu, Oct 25, 2007 at 09:17:54AM +0200, Pierre Muller wrote:
> >   The next question is for Eli, as DJGPP maintainer:
> >   I have locally a working version of doscheck,
> > which avoid duplicates in different directories,
> > but still reveals some conflicts for DJGPP.
> >   The most notable is
> > gdb/ChangeLog and gdb/ChangeLog-2006
> > but there are others in bfd, libdecnumber, gdb/testsuite
> > and other directories,
> 
> This means you should probably update config/djgpp/fnchange.lst.

Yes.  Any files that don't have a mapping in fnchange.lst and show in
the "conflicts" part of `doschk's output should be added to
fnchange.lst.

Re: [RFA] gdb_ari.sh changes to discard some file types for specific tests

[Eli Zaretskii]

> From: "Pierre Muller" <muller@ics.u-strasbg.fr>
> Cc: <gdb-patches@sourceware.org>
> Date: Thu, 25 Oct 2007 09:17:54 +0200
> 
>   I changed the output of doscheck to
> avoid getting false duplicates for files
> in different directories.

What are those false duplicates? can you show an example?

I use doschk quite frequently, but never met such false duplicates.

>   By the way, is the script that
> generates the DJGPP gdb source public?
> If yes, where can I find it?

The only DJGPP-specific script required to build GDB is
gdb/config/djgpp/djconfig.sh.  You are supposed to run it instead of
the standard ./configure (djconfig.sh will run the top-level configure
script as part of its job).  The rest is just like on any other
system: run Make and watch it build (hopefully; I haven't built it for
quite some time, due to lack of free time).

RE: [RFA] gdb_ari.sh changes to discard some file types for specific tests

[Pierre Muller]

> -----Original Message-----
> From: Eli Zaretskii [mailto:eliz@gnu.org]
> Sent: Thursday, October 25, 2007 11:55 PM
> To: Pierre Muller
> Cc: drow@false.org; gdb-patches@sourceware.org
> Subject: Re: [RFA] gdb_ari.sh changes to discard some file types for
> specific tests
> 
> > From: "Pierre Muller" <muller@ics.u-strasbg.fr>
> > Cc: <gdb-patches@sourceware.org>
> > Date: Thu, 25 Oct 2007 09:17:54 +0200
> >
> >   I changed the output of doscheck to
> > avoid getting false duplicates for files
> > in different directories.
> 
> What are those false duplicates? can you show an example?
> 
> I use doschk quite frequently, but never met such false duplicates.

  The errors are not in doscheck itself, but in the
way the resulting data is handled in update_web_ari
which will put together duplicates like
  gdb/ChangeLog and gdb/ChangeLog-2006
and
  opcodes/ChangeLog and opcodes/ChangeLog-2006
into a single CHANGELO entry.

  I will send the changes in a separate email.

> >   By the way, is the script that
> > generates the DJGPP gdb source public?
> > If yes, where can I find it?
> 
> The only DJGPP-specific script required to build GDB is
> gdb/config/djgpp/djconfig.sh.  You are supposed to run it instead of
> the standard ./configure (djconfig.sh will run the top-level configure
> script as part of its job).  The rest is just like on any other
> system: run Make and watch it build (hopefully; I haven't built it for
> quite some time, due to lack of free time).

  No, the question was:
how are the zip files for the sources of the DJGPP port of
GDB produced?

  They do not include all the directories
that are in the official sources, but I was wondering
it the zip was produced manually or
if there was a special script for that, that would
remove lots of unnecessary directories.
  If this script exists, it would be useful to
limit the scope of doscheck.c to these directories
included in the final source zip for DJGPP.

Pierre

Re: [RFA] gdb_ari.sh changes to discard some file types for specific tests

[Eli Zaretskii]

> From: "Pierre Muller" <muller@ics.u-strasbg.fr>
> Cc: <drow@false.org>, 	<gdb-patches@sourceware.org>
> Date: Fri, 26 Oct 2007 09:19:12 +0200
> 
> > > From: "Pierre Muller" <muller@ics.u-strasbg.fr>
> > > Cc: <gdb-patches@sourceware.org>
> > > Date: Thu, 25 Oct 2007 09:17:54 +0200
> > >
> > >   I changed the output of doscheck to
> > > avoid getting false duplicates for files
> > > in different directories.
> > 
> > What are those false duplicates? can you show an example?
> > 
> > I use doschk quite frequently, but never met such false duplicates.
> 
>   The errors are not in doscheck itself, but in the
> way the resulting data is handled in update_web_ari

Well, you did say that you fixed doschk...

> which will put together duplicates like
>   gdb/ChangeLog and gdb/ChangeLog-2006
> and
>   opcodes/ChangeLog and opcodes/ChangeLog-2006
> into a single CHANGELO entry.

That's true, but so what?  The output clearly states the directory of
each file, doesn't it?

>   No, the question was:
> how are the zip files for the sources of the DJGPP port of
> GDB produced?

Ah, sorry for misunderstanding.  The omission of source directories is
manual.