[RFA] Fix "show convenience" test

[RFA] Fix "show convenience" test

[Sérgio Durigan Júnior]

[-- Attachment #1: Type: Text/Plain, Size: 1093 bytes --]

Hi guys,

This is a simple one.  After running GDB's testsuite, I saw that one of the 
gdb.base/default.exp's tests was not passing.  It tested the output of the 
"show convenience" command.

After a little investigation, it seems that:

http://sourceware.org/ml/gdb-patches/2009-02/msg00165.html

Is the responsible for this, since this patch was creating an internal 
convenience variable ($_siginfo) which appears in the "show convenience" 
command, while the original test expects to receive a warning saying that no 
convenience variable is created at the moment.  If I understood the patch 
correctly, the $_siginfo will be always created and therefore the test should 
now expect "$_siginfo = void" instead of this warning.  Am I missing 
something?

Regards,

-- 
Sérgio Durigan Júnior
Linux on Power Toolchain - Software Engineer
Linux Technology Center - LTC
IBM Brazil

gdb/testsuite/ChangeLog:

2009-09-15  Sergio Durigan Junior <sergiosdj@gmail.com>

	* gdb.base/default.exp: Updated the "show convenience" test
	in order to accept the new output.

[-- Attachment #2: show-convenience.patch --]
[-- Type: text/x-patch, Size: 900 bytes --]

diff --git a/gdb/testsuite/gdb.base/default.exp b/gdb/testsuite/gdb.base/default.exp
index a223982..a0cb11e 100644
--- a/gdb/testsuite/gdb.base/default.exp
+++ b/gdb/testsuite/gdb.base/default.exp
@@ -598,7 +598,7 @@ gdb_test "show complaints" "Max number of complaints about incorrect symbols is
 #test show confirm
 gdb_test "show confirm" "Whether to confirm potentially dangerous operations is o\[a-z\]*." "show confirm"
 #test show convenience
-gdb_test "show convenience" "No debugger convenience variables now defined.(\[^\r\n\]*\[\r\n\])+Convenience variables have names starting with \".\";(\[^\r\n\]*\[\r\n\])+use \"set\" as in \"set .foo = 5\" to define them." "show convenience"
+gdb_test "show convenience" "\\\$_siginfo = void" "show convenience"
 #test show directories
 gdb_test "show directories" "Source directories searched: .cdir\[:;\].cwd" "show directories"
 #test show editing

Re: [RFA] Fix "show convenience" test

[Pedro Alves]

On Wednesday 16 September 2009 01:32:15, Sérgio Durigan Júnior wrote:

> This is a simple one.  After running GDB's testsuite, I saw that one of the 
> gdb.base/default.exp's tests was not passing.  It tested the output of the 
> "show convenience" command.
> 
> After a little investigation, it seems that:
> 
> http://sourceware.org/ml/gdb-patches/2009-02/msg00165.html
> 

Hmm, I had already approved Paul's similar follow up patch at
<http://sourceware.org/ml/gdb-patches/2009-04/msg00105.html>.
I looks like it was waiting for a docs review.  As before, this is
fine with me.

-- 
Pedro Alves

Re: [RFA] Fix "show convenience" test

[Eli Zaretskii]

> From: Pedro Alves <pedro@codesourcery.com>
> Date: Wed, 16 Sep 2009 14:16:22 +0100
> Cc: gdb-patches@sourceware.org
> 
> Hmm, I had already approved Paul's similar follow up patch at
> <http://sourceware.org/ml/gdb-patches/2009-04/msg00105.html>.
> I looks like it was waiting for a docs review.

Ouch! sorry about that (binary attachments tend to have this effect on
me ;-).

>  The variable @code{$_siginfo} is bound to extra signal information
> -inspection (@pxref{extra signal information}).
> +inspection (@pxref{extra signal information}). Note that @code{$_siginfo}
> +could be empty. For example, it will be empty before you execute the
> +@code{run} command.
>  @end table

A couple of comments:

  . "the variable ... is bound to extra signal information inspection"
    is awkward wording.  "How about "contains extra signal
    information"?

  . I don't understand the note about it being ``empty''?  What does
    that mean, exactly, and why the example is talking about before
    `run'?  I could understand that it's empty if no signals were
    delivered yet, but what does `run' has to do with this?

  . Please leave two spaces between sentences, not one.

Thanks.

Re: [RFA] Fix "show convenience" test

[Paul Pluzhnikov]

[-- Attachment #1: Type: text/plain, Size: 3127 bytes --]

On Wed, Sep 16, 2009 at 10:08 AM, Eli Zaretskii <eliz@gnu.org> wrote:

>> From: Pedro Alves <pedro@codesourcery.com>
>> Date: Wed, 16 Sep 2009 14:16:22 +0100
>> Cc: gdb-patches@sourceware.org
>>
>> Hmm, I had already approved Paul's similar follow up patch at
>> <http://sourceware.org/ml/gdb-patches/2009-04/msg00105.html>.

I now see that you did. I lost that fact in the ensuing discussion of
testsuite failures :-(

>> I looks like it was waiting for a docs review.
>
> Ouch! sorry about that (binary attachments tend to have this effect on
> me ;-).

The attachment was actually text, just base64-encoded:
Content-Type: text/plain; charset=US-ASCII; name="gdb-convenience.20090406.txt"

Unfortunately, my attempts to convince gmail not to do that have failed :-(

>>  The variable @code{$_siginfo} is bound to extra signal information
>> -inspection (@pxref{extra signal information}).
>> +inspection (@pxref{extra signal information}). Note that @code{$_siginfo}
>> +could be empty. For example, it will be empty before you execute the
>> +@code{run} command.
>>  @end table
>
> A couple of comments:
>
>  . "the variable ... is bound to extra signal information inspection"
>    is awkward wording.  "How about "contains extra signal
>    information"?

Fixed.

>  . I don't understand the note about it being ``empty''?  What does
>    that mean, exactly, and why the example is talking about before
>    `run'?  I could understand that it's empty if no signals were
>    delivered yet, but what does `run' has to do with this?

Before the first 'run' is one rare case where the application has not yet
received any signals. I think this example could be simply removed
without much loss. Let me know if you'd like me to do that.

>
>  . Please leave two spaces between sentences, not one.

Sorry about that.  Fixed.


How does the attached patch look?

Here's the doc part, so you don't have to deal with base64:

Index: doc/gdb.texinfo
===================================================================
RCS file: /cvs/src/src/gdb/doc/gdb.texinfo,v
retrieving revision 1.624
diff -u -p -u -r1.624 gdb.texinfo
--- doc/gdb.texinfo	15 Sep 2009 18:51:25 -0000	1.624
+++ doc/gdb.texinfo	17 Sep 2009 16:33:19 -0000
@@ -7819,8 +7819,10 @@ the program being debugged terminates.

 @item $_siginfo
 @vindex $_siginfo@r{, convenience variable}
-The variable @code{$_siginfo} is bound to extra signal information
-inspection (@pxref{extra signal information}).
+The variable @code{$_siginfo} contains extra signal information
+(@pxref{extra signal information}).  Note that @code{$_siginfo}
+could be empty, if the application has not yet received any signals.
+For example, it will be empty before you execute the @code{run} command.
 @end table

 On HP-UX systems, if you refer to a function or variable name that


Thanks,
-- 
Paul Pluzhnikov


testsuite/ChangeLog
2009-09-17  Paul Pluzhnikov  <ppluzhnikov@google.com>

	    * gdb.base/default.exp: Fix "show convenience".

doc/ChangeLog
2009-09-17  Paul Pluzhnikov  <ppluzhnikov@google.com>

	    * gdb.texinfo (convenince variables): Mention
	    $_siginfo could be empty.

[-- Attachment #2: gdb-convenience.20090917.txt --]
[-- Type: text/plain, Size: 1911 bytes --]

Index: testsuite/gdb.base/default.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.base/default.exp,v
retrieving revision 1.31
diff -u -p -u -r1.31 default.exp
--- testsuite/gdb.base/default.exp	13 Aug 2009 14:58:27 -0000	1.31
+++ testsuite/gdb.base/default.exp	17 Sep 2009 16:33:19 -0000
@@ -598,7 +598,7 @@ gdb_test "show complaints" "Max number o
 #test show confirm
 gdb_test "show confirm" "Whether to confirm potentially dangerous operations is o\[a-z\]*." "show confirm"
 #test show convenience
-gdb_test "show convenience" "No debugger convenience variables now defined.(\[^\r\n\]*\[\r\n\])+Convenience variables have names starting with \".\";(\[^\r\n\]*\[\r\n\])+use \"set\" as in \"set .foo = 5\" to define them." "show convenience"
+gdb_test "show convenience" "\\\$_siginfo = void" "show convenience"
 #test show directories
 gdb_test "show directories" "Source directories searched: .cdir\[:;\].cwd" "show directories"
 #test show editing
Index: doc/gdb.texinfo
===================================================================
RCS file: /cvs/src/src/gdb/doc/gdb.texinfo,v
retrieving revision 1.624
diff -u -p -u -r1.624 gdb.texinfo
--- doc/gdb.texinfo	15 Sep 2009 18:51:25 -0000	1.624
+++ doc/gdb.texinfo	17 Sep 2009 16:33:19 -0000
@@ -7819,8 +7819,10 @@ the program being debugged terminates.
 
 @item $_siginfo
 @vindex $_siginfo@r{, convenience variable}
-The variable @code{$_siginfo} is bound to extra signal information
-inspection (@pxref{extra signal information}).
+The variable @code{$_siginfo} contains extra signal information
+(@pxref{extra signal information}).  Note that @code{$_siginfo}
+could be empty, if the application has not yet received any signals.
+For example, it will be empty before you execute the @code{run} command.
 @end table
 
 On HP-UX systems, if you refer to a function or variable name that

Re: [RFA] Fix "show convenience" test

[Eli Zaretskii]

> Date: Thu, 17 Sep 2009 09:37:08 -0700
> From: Paul Pluzhnikov <ppluzhnikov@google.com>
> Cc: Pedro Alves <pedro@codesourcery.com>, sergiodj@linux.vnet.ibm.com,
>         gdb-patches@sourceware.org
> 
> > Ouch! sorry about that (binary attachments tend to have this effect on
> > me ;-).
> 
> The attachment was actually text, just base64-encoded:

Lately, my eyes lost their ability to read base64 without help ;-)

> How does the attached patch look?

It's fine now, thanks.

> Here's the doc part, so you don't have to deal with base64:

Thanks for making it easier for me.

Re: [RFA] Fix "show convenience" test

[Paul Pluzhnikov]

On Thu, Sep 17, 2009 at 10:26 AM, Eli Zaretskii <eliz@gnu.org> wrote:

>> The attachment was actually text, just base64-encoded:
>
> Lately, my eyes lost their ability to read base64 without help ;-)

You appear to be using emacs; surely it can help?

I've used VM in the past, and had no problems decoding base64 (or anything
else for that matter).

BTW, here is a response from gmail folks to my request to be able to send
text/plain attachments.

<quote>

  *sigh*

  How old is base64 now, has it hit 15 years yet?  "text based mail reader"
  my left foot, there's nothing to prevent a text based mail reader from
  decoding mail in a 15 year old standard.

  I can accept the argument that a pure text, 7bit ascii message of a
  text subtype shouldn't be encoded automatically.  I'm even suprised
  that its base64 encoded and not qp-encoded if there happened to be
  line-ending/wrapping concerns.

</quote>

>> How does the attached patch look?
>
> It's fine now, thanks.

So committed.

Thanks,
-- 
Paul Pluzhnikov

Re: [RFA] Fix "show convenience" test

[Pedro Alves]

On Thursday 17 September 2009 18:57:31, Paul Pluzhnikov wrote:
> <quote>
> 
>   *sigh*
> 
>   How old is base64 now, has it hit 15 years yet?  "text based mail reader"
>   my left foot, there's nothing to prevent a text based mail reader from
>   decoding mail in a 15 year old standard.
> 
>   I can accept the argument that a pure text, 7bit ascii message of a
>   text subtype shouldn't be encoded automatically.  I'm even suprised
>   that its base64 encoded and not qp-encoded if there happened to be
>   line-ending/wrapping concerns.
> 

I hope that means that they'll address it.  :-)  This also prevents
mail archives from showing the patch inline.  Compare:

 http://sourceware.org/ml/gdb-patches/2009-04/msg00105.html
 http://sourceware.org/cgi-bin/get-raw-msg?listname=gdb-patches&date=2009-04&msgid=8ac60eac0904061159v4deb9d48n1cf791463e587e54%40mail.gmail.com

vs

 http://sourceware.org/ml/gdb-patches/2009-09/msg00496.html
 http://sourceware.org/cgi-bin/get-raw-msg?listname=gdb-patches&date=2009-09&msgid=200909152132.15317.sergiodj%40linux.vnet.ibm.com

-- 
Pedro Alves

Re: [RFA] Fix "show convenience" test

[Eli Zaretskii]

> Date: Thu, 17 Sep 2009 10:57:31 -0700
> From: Paul Pluzhnikov <ppluzhnikov@google.com>
> Cc: pedro@codesourcery.com, sergiodj@linux.vnet.ibm.com,
>         gdb-patches@sourceware.org
> 
> On Thu, Sep 17, 2009 at 10:26 AM, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> >> The attachment was actually text, just base64-encoded:
> >
> > Lately, my eyes lost their ability to read base64 without help ;-)
> 
> You appear to be using emacs; surely it can help?

Sure, but it needs to be asked nicely ;-)

> I've used VM in the past, and had no problems decoding base64 (or anything
> else for that matter).

There are no problems to do that in any of the Emacs MUAs, except that
in my setup this is done only upon request.  Imagine me scrolling
through 100-odd mails, and you will understand why I sometimes let the
encoded ones woosh by.

>   How old is base64 now, has it hit 15 years yet?  "text based mail reader"
>   my left foot, there's nothing to prevent a text based mail reader from
>   decoding mail in a 15 year old standard.

There's that thing called ``viri'' and ``Trojans''.  I guess they
never heard about them.

Re: [RFA] Fix "show convenience" test

[Paul Pluzhnikov]

On Thu, Sep 17, 2009 at 11:16 AM, Pedro Alves <pedro@codesourcery.com> wrote:

> I hope that means that they'll address it.  :-)

I don't have much hope for that :-(

> This also prevents
> mail archives from showing the patch inline.  Compare:
>
> http://sourceware.org/ml/gdb-patches/2009-04/msg00105.html
> http://sourceware.org/cgi-bin/get-raw-msg?listname=gdb-patches&date=2009-04&msgid=8ac60eac0904061159v4deb9d48n1cf791463e587e54%40mail.gmail.com
>
> vs
>
> http://sourceware.org/ml/gdb-patches/2009-09/msg00496.html
> http://sourceware.org/cgi-bin/get-raw-msg?listname=gdb-patches&date=2009-09&msgid=200909152132.15317.sergiodj%40linux.vnet.ibm.com

Yes, but it's only 1 click away.


-- 
Paul Pluzhnikov

Re: [RFA] Fix "show convenience" test

[Joseph S. Myers]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 943 bytes --]

On Thu, 17 Sep 2009, Pedro Alves wrote:

> On Thursday 17 September 2009 18:57:31, Paul Pluzhnikov wrote:
> > <quote>
> > 
> >   *sigh*
> > 
> >   How old is base64 now, has it hit 15 years yet?  "text based mail reader"
> >   my left foot, there's nothing to prevent a text based mail reader from
> >   decoding mail in a 15 year old standard.
> > 
> >   I can accept the argument that a pure text, 7bit ascii message of a
> >   text subtype shouldn't be encoded automatically.  I'm even suprised
> >   that its base64 encoded and not qp-encoded if there happened to be
> >   line-ending/wrapping concerns.
> > 
> 
> I hope that means that they'll address it.  :-)  This also prevents
> mail archives from showing the patch inline.  Compare:

It's a bug in the mail archiving software if it decides that based on 
Content-Transfer-Encoding rather than Content-Disposition.

-- 
Joseph S. Myers
joseph@codesourcery.com

Re: [RFA] Fix "show convenience" test

[Pedro Alves]

On Thursday 17 September 2009 20:25:01, Joseph S. Myers wrote:

> > I hope that means that they'll address it.  :-)  This also prevents
> > mail archives from showing the patch inline.  Compare:
> 
> It's a bug in the mail archiving software if it decides that based on 
> Content-Transfer-Encoding rather than Content-Disposition.

I don't know the mail archiving software's algorithm, but I
sort of disagree, while I do get your point.  Rather, I'd say that
it's a feature that "Content-Disposition: attachment" +
"Content-Transfer-Encoding: 7bit" is displayed inline.

-- 
Pedro Alves