Produces better error messages when programmers forget to document a property.

Produces better error messages when programmers forget to document a property.

[MikeSol, 2011-07-20 12:47:56]

I've needed something like this for a long time - lemme know what you think!

Cheers,
MS

[Neil Puttock, 2011-07-23 21:03:36]

LGTM.

[Janek Warchol, 2011-07-23 21:07:34]

On 2011/07/20 12:47:56, MikeSol wrote:
> I've needed something like this for a long time - lemme know what you think!

I don't understand how it works, but from what i suppose it's results will be
very nice!

thanks,
Janek

[pkx166h, 2011-07-23 22:05:34]

Can you give me an idea what his does and how to test this or what I am going to
see as someone who runs a lot of make/reg tests?

[Neil Puttock, 2011-07-23 22:24:57]

On 23 July 2011 23:05,  <pkx166h@gmail.com> wrote:
> Can you give me an idea what his does and how to test this or what I am
> going to see as someone who runs a lot of make/reg tests?

If somebody forgets to document a new property (in
scm/define-grob-properties.scm or scm/define-context-properties.scm)
this will ensure the internals documentation generation aborts with a
useful error message.  If you want to test it, apply Mike's patch then
try the following bad patch.  It adds an undocumented grob property
called `foo' to the accidental-interface.

Cheers,
Neil

[Neil Puttock, 2011-07-23 22:25:58]

http://codereview.appspot.com/4801045/diff/1/scm/documentation-lib.scm
File scm/documentation-lib.scm (right):

http://codereview.appspot.com/4801045/diff/1/scm/documentation-lib.scm#newcod...
scm/documentation-lib.scm:184: (ly:error (_ "cannot find description for
property ~S (~S)") sym where))
property `~S'

[Graham Percival, 2011-07-24 01:43:50]

On Sat, Jul 23, 2011 at 10:05:34PM +0000, pkx166h@gmail.com wrote:
> Can you give me an idea what his does and how to test this or what I am
> going to see as someone who runs a lot of make/reg tests?

To initial acceptance of this patch:
- you should see no change in the regtests.

After it's pushed, and you're testing other patches:
- you might see some extra lines in the regtest log files.  Those
  should be considered as serious warnings, and a patch which adds
  those warnings should be rejected.

Cheers,
- Graham

[mike_apollinemike.com, 2011-07-24 08:41:27]

On Jul 24, 2011, at 12:05 AM, pkx166h@gmail.com wrote:

> Can you give me an idea what his does and how to test this or what I am
> going to see as someone who runs a lot of make/reg tests?
> 

Try adding an undocumented property to an interface (i.e. add "foo" to the Stem
interface) and then run make.  Before, make would throw an error about
procedure-name when building internals.texi.  Now, it will still throw that
error, but before that, it'll explicitly mention that "foo" is undocumented,
allowing for one to then document it.

Cheers,
MS

[mike_apollinemike.com, 2011-07-24 08:42:00]

On Jul 24, 2011, at 12:24 AM, Neil Puttock wrote:

> On 23 July 2011 23:05,  <pkx166h@gmail.com> wrote:
>> Can you give me an idea what his does and how to test this or what I am
>> going to see as someone who runs a lot of make/reg tests?
> 
> If somebody forgets to document a new property (in
> scm/define-grob-properties.scm or scm/define-context-properties.scm)
> this will ensure the internals documentation generation aborts with a
> useful error message.  If you want to test it, apply Mike's patch then
> try the following bad patch.  It adds an undocumented grob property
> called `foo' to the accidental-interface.
> 
> Cheers,
> Neil


Ha...LilyPond minds think alike...

Cheers,
MS

[James.Lowe_datacore.com, 2011-07-25 11:18:28]

Hello,

)-----Original Message-----
)From: lilypond-devel-bounces+james.lowe=datacore.com@gnu.org
)[mailto:lilypond-devel-bounces+james.lowe=datacore.com@gnu.org] On
)Behalf Of mike@apollinemike.com
)Sent: 24 July 2011 09:41
)To: mtsolo@gmail.com; n.puttock@gmail.com;
)lemniskata.bernoullego@gmail.com; pkx166h@gmail.com; lilypond-
)devel@gnu.org; reply@codereview.appspotmail.com
)Subject: Re: Produces better error messages when programmers forget
)to document a property. (issue4801045)
)
)On Jul 24, 2011, at 12:05 AM, pkx166h@gmail.com wrote:
)
)> Can you give me an idea what his does and how to test this or what I
)> am going to see as someone who runs a lot of make/reg tests?
)>
)
)Try adding an undocumented property to an interface (i.e. add "foo" to
)the Stem interface) and then run make.  Before, make would throw an
)error about procedure-name when building internals.texi.  Now, it will still
)throw that error, but before that, it'll explicitly mention that "foo" is
)undocumented, allowing for one to then document it.
)
)Cheers,
)MS

Thanks, for this information. I'll try to run the test tonight - I ran out of
time yesterday - so I can see what it looks like when it does what it is
supposed to do, using Neil's 'broken' patch.

James

[MikeSol, 2011-07-27 08:43:05]

On 2011/07/25 11:18:28, James.Lowe_datacore.com wrote:
> Hello,
> 
> )-----Original Message-----
> )From: mailto:lilypond-devel-bounces+james.lowe=datacore.com@gnu.org
> )[mailto:lilypond-devel-bounces+james.lowe=datacore.com@gnu.org] On
> )Behalf Of mailto:mike@apollinemike.com
> )Sent: 24 July 2011 09:41
> )To: mtsolo@gmail.com; n.puttock@gmail.com;
> )lemniskata.bernoullego@gmail.com; pkx166h@gmail.com; lilypond-
> )devel@gnu.org; mailto:reply@codereview.appspotmail.com
> )Subject: Re: Produces better error messages when programmers forget
> )to document a property. (issue4801045)
> )
> )On Jul 24, 2011, at 12:05 AM, mailto:pkx166h@gmail.com wrote:
> )
> )> Can you give me an idea what his does and how to test this or what I
> )> am going to see as someone who runs a lot of make/reg tests?
> )>
> )
> )Try adding an undocumented property to an interface (i.e. add "foo" to
> )the Stem interface) and then run make.  Before, make would throw an
> )error about procedure-name when building internals.texi.  Now, it will still
> )throw that error, but before that, it'll explicitly mention that "foo" is
> )undocumented, allowing for one to then document it.
> )
> )Cheers,
> )MS
> 
> Thanks, for this information. I'll try to run the test tonight - I ran out of
> time yesterday - so I can see what it looks like when it does what it is
> supposed to do, using Neil's 'broken' patch.
> 
> James

Pushed as 922d59a242646c2488738d77263580427bd7764b.

Cheers,
MS

[James.Lowe_datacore.com, 2011-07-27 10:14:26]

Hello,

)-----Original Message-----
)From: mtsolo@gmail.com [mailto:mtsolo@gmail.com]
)Sent: 27 July 2011 09:43
)To: n.puttock@gmail.com; lemniskata.bernoullego@gmail.com;
)pkx166h@gmail.com; graham@percival-music.ca;
)mike@apollinemike.com; James Lowe
)Cc: lilypond-devel@gnu.org; reply@codereview.appspotmail.com
)Subject: Re: Produces better error messages when programmers forget
)to document a property. (issue4801045)
)
)On 2011/07/25 11:18:28, James.Lowe_datacore.com wrote:
)> Hello,
)
)> )-----Original Message-----
)> )From: mailto:lilypond-devel-
)bounces+james.lowe=datacore.com@gnu.org
)> )[mailto:lilypond-devel-bounces+james.lowe=datacore.com@gnu.org]
)On
)> )Behalf Of mailto:mike@apollinemike.com
)> )Sent: 24 July 2011 09:41
)> )To: mtsolo@gmail.com; n.puttock@gmail.com;
)> )lemniskata.bernoullego@gmail.com; pkx166h@gmail.com; lilypond-
)> )devel@gnu.org; mailto:reply@codereview.appspotmail.com
)> )Subject: Re: Produces better error messages when programmers
)forget
)> )to document a property. (issue4801045)
)> )
)> )On Jul 24, 2011, at 12:05 AM, mailto:pkx166h@gmail.com wrote:
)> )
)> )> Can you give me an idea what his does and how to test this or what
)I
)> )> am going to see as someone who runs a lot of make/reg tests?
)> )>
)> )
)> )Try adding an undocumented property to an interface (i.e. add "foo"
)to
)> )the Stem interface) and then run make.  Before, make would throw an
)> )error about procedure-name when building internals.texi.  Now, it
)will still
)> )throw that error, but before that, it'll explicitly mention that
)"foo" is
)> )undocumented, allowing for one to then document it.
)> )
)> )Cheers,
)> )MS
)
)> Thanks, for this information. I'll try to run the test tonight - I ran
)out of
)> time yesterday - so I can see what it looks like when it does what it
)is
)> supposed to do, using Neil's 'broken' patch.
)
)> James
)
)Pushed as 922d59a242646c2488738d77263580427bd7764b.
)
)Cheers,
)MS
)
)http://codereview.appspot.com/4801045/

Oh.. Didn't get a chance to test it yet :)

But I have Neil's broken patch so I can still test it tonight for my own
personal point of view.

James