Re: [PATCH][TEST-CASE][DOC] Implementation of pipe to pass GDB's command output to the shell.

[Abhijit Halder <abhijit dot k dot halder at gmail dot com>]

On Tue, Aug 30, 2011 at 11:36 PM, Eli Zaretskii <eliz@gnu.org> wrote:
>> Date: Tue, 30 Aug 2011 22:08:22 +0530
>> From: Abhijit Halder <abhijit.k.halder@gmail.com>
>> Cc: Eli Zaretskii <eliz@gnu.org>, gdb-patches@sourceware.org, sergiodj@redhat.com,
>> ? ? ? jan.kratochvil@redhat.com, tromey@redhat.com
>>
>> + ? ? * gdb.texinfo (pipe command): New node.
>
> You didn't write a new node, you just added text to the existing node.
> So you should state the name of that node in the parentheses.
>
Sorry, I missed editing this part.

>> +If you want to process the output of a @value{GDBN} command using some shell
>> +command or some script, that can be done by using the command @code{pipe}. ?You
>> +don't even need to get a shell for that. ?The same could be achieved by
>> +enabling logging and redirecting the output of @value{GDBN} command to some
>> +file and then processing the file content by invoking a shell using
>> +@code{shell} command. ?But this incurs higher costs as the explicit invokation
>> +of shell involves execution of its startup scripts and locating the file
>> +containing @value{GDBN}'s command output involves directory lookup; even though
>> +OS may optimize these processes by its cache mechanism.
>
> I would drop this part, and leave only the first sentence. ?We don't
> need to justify commands in the manual.
>
>> +@table @code
>> +@item pipe @var{dlim} @var{gdbcmd} @var{dlim} @var{shellcmd}
>
> You need a "@kindex pipe" here, to index this command as we do with
> all other commands.
>
>> +@var{dlim} is a string of arbitrary length, containing no whitespace and no
>> +leading '-', acts as a separator between a @value{GDBN} command @var{gdbcmd}
> ? ? ? ? ? ^^^
> @samp{-}
>
>> +In the above example @samp{@var{<br>}} acts as a delimiter. ?The output of
>
> Just @samp{<br>}, without @var.
>
>> +@samp{print dd_tbl} is passed to the shell command @samp{sed 's/@}/\n/g' | @
>> +grep ".test_dd" | tr ',' '\n'} for processing.
>
> I would suggest to add that the example does not use `|' as a
> delimiter to allow the shell command to use it in its usual shell
> semantics.
>
In the above example one can use `|' as the delimiter. it will not
break anything syntactically but will not be readable much, hence I
avoided using `|' as delimiter. Do we need to mention that here?

>
>> +In the given example the output of @value{GDBN} command is huge and not well
>> +formated. ?The use of shell commands like ``sed'', ``tr'' and ``grep'' ease
>
> Please use @command{sed}, @command{tr}, etc.
>
> Thanks.
>

I made the corrections as suggested. Please review this:

Thanks,
Abhijit Halder

diff -rup src/gdb/doc/ChangeLog dst/gdb/doc/ChangeLog
--- src/gdb/doc/ChangeLog	2011-08-27 20:35:38.459934029 +0530
+++ dst/gdb/doc/ChangeLog	2011-08-31 13:33:59.188118016 +0530
@@ -1,3 +1,7 @@
+2011-08-29  Abhijit Halder  <abhijit.k.halder@gmail.com>
+
+	* gdb.texinfo (Shell Commands): Add pipe command.
+
 2011-08-25  Andrew Oakley  <andrew@ado.is-a-geek.net>
 
 	* gdb.texinfo (Types In Python): Document 'bitpos' for enums.
diff -rup src/gdb/doc/gdb.texinfo dst/gdb/doc/gdb.texinfo
--- src/gdb/doc/gdb.texinfo	2011-08-26 21:11:39.938354007 +0530
+++ dst/gdb/doc/gdb.texinfo	2011-08-31 13:43:45.836117888 +0530
@@ -1367,6 +1367,48 @@ Execute the @code{make} program with the
 arguments.  This is equivalent to @samp{shell make @var{make-args}}.
 @end table
 
+If you want to process the output of a @value{GDBN} command using some shell
+command or some script, that can be done by using the command @code{pipe}.
+
+@table @code
+@kindex pipe
+@item pipe @var{dlim} @var{gdbcmd} @var{dlim} @var{shellcmd}
+@var{dlim} is a string of arbitrary length, containing no whitespace and no
+leading @samp{-}, acts as a separator between a @value{GDBN} command
+@var{gdbcmd} and a shell command @var{shellcmd}.  The shell command should be
+in compliance with the syntax of the default shell.
+@end table
+
+@smallexample
+(@value{GDBP}) @b{ptype dd_tbl}
+type = struct dd @{
+    int dd_handle;
+    const char *dd_name;
+    int dd_major;
+    int dd_minor;
+    void *dd_code;
+    void *dd_data;
+@} [1024]
+(@value{GDBP}) @b{pipe <br> print dd_tbl <br> sed 's/@}/\n/g' | grep @
+"\.test_dd" | tr ',' '\n'}
+
+  @{dd_handle = 10
+  dd_name = 0x8048538 ".test_dd"
+  dd_major = 100
+  dd_minor = 0
+  dd_code = 0xcc
+  dd_data = 0x80
+@end smallexample
+
+In the above example @samp{<br>} acts as a delimiter.  The output of
+@samp{print dd_tbl} is passed to the shell command @samp{sed 's/@}/\n/g' | @
+grep ".test_dd" | tr ',' '\n'} for processing.
+
+In the given example the output of @value{GDBN} command @samp{print dd_tbl} is
+huge and not well formated.  The use of shell commands like @command{sed},
+@command{tr} and @command{grep} ease the searching of desired pattern and hence
+ease debugging.
+
 @node Logging Output
 @section Logging Output
 @cindex logging @value{GDBN} output