Documentation of kconfig language differs from implementation regarding existence of symbols

Documentation of kconfig language differs from implementation regarding existence of symbols

[Martin Walch]

[-- Attachment #1.1: Type: text/plain, Size: 1049 bytes --]

The file linux/Documentation/kbuild/kconfig-language.txt says:

> The following two methods produce the same kconfig symbol dependencies
> but differ greatly in kconfig symbol existence (production) in the
> generated config file.
>
> case 1:
>
> config FOO
>         tristate "about foo"
>         depends on BAR
>
> vs. case 2:
>
> if BAR
> config FOO
>         tristate "about foo"
> endif
>
> In case 1, the symbol FOO will always exist in the config file (given
> no other dependencies).  In case 2, the symbol FOO will only exist in
> the config file if BAR is enabled.

However, I can not reproduce this. The attached file contains both cases. When 
running make menuconfig, setting BAR0 and BAR1 both to n, and saving the 
configuration, there is neither FOO0 nor FOO1 in the resulting configuration 
file.

According to the documentation, at least FOO0 should exist, even if BAR0 is 
set to n. But this is not the case. AFAICS, both versions behave equivalently. 
I suggest changing the documentation accordingly.

Regards
Martin Walch
-- 

[-- Attachment #1.2: FOO.kconf --]
[-- Type: text/plain, Size: 187 bytes --]

config FOO0
        tristate "FOO0"
        depends on BAR0

config BAR0
        tristate "BAR0"

config BAR1
        tristate "BAR1"

if BAR1

config FOO1
        tristate "FOO1"

endif

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 490 bytes --]

Re: Documentation of kconfig language differs from implementation regarding existence of symbols

[Michal Marek]

Dne 10.10.2012 02:17, Martin Walch napsal(a):
> The file linux/Documentation/kbuild/kconfig-language.txt says:
> 
>> The following two methods produce the same kconfig symbol
>> dependencies but differ greatly in kconfig symbol existence
>> (production) in the generated config file.
>> 
>> case 1:
>> 
>> config FOO tristate "about foo" depends on BAR
>> 
>> vs. case 2:
>> 
>> if BAR config FOO tristate "about foo" endif
>> 
>> In case 1, the symbol FOO will always exist in the config file
>> (given no other dependencies).  In case 2, the symbol FOO will
>> only exist in the config file if BAR is enabled.
> 
> However, I can not reproduce this. The attached file contains both
> cases. When running make menuconfig, setting BAR0 and BAR1 both to
> n, and saving the configuration, there is neither FOO0 nor FOO1 in
> the resulting configuration file.

Indeed, that paragraph seems wrong. I think we should revert commit
64b81ed. Randy, Arnaud?

Thanks,
Michal

Re: Documentation of kconfig language differs from implementation regarding existence of symbols

[Yann E. MORIN]

Michal, Martin, All,

On Thursday 11 October 2012 22:27:36 Michal Marek wrote:
> Dne 10.10.2012 02:17, Martin Walch napsal(a):
> > The file linux/Documentation/kbuild/kconfig-language.txt says:
> > 
> >> The following two methods produce the same kconfig symbol
> >> dependencies but differ greatly in kconfig symbol existence
> >> (production) in the generated config file.
> >> 
> >> case 1:
> >> 
> >> config FOO tristate "about foo" depends on BAR
> >> 
> >> vs. case 2:
> >> 
> >> if BAR config FOO tristate "about foo" endif
> >> 
> >> In case 1, the symbol FOO will always exist in the config file
> >> (given no other dependencies).  In case 2, the symbol FOO will
> >> only exist in the config file if BAR is enabled.
> > 
> > However, I can not reproduce this. The attached file contains both
> > cases. When running make menuconfig, setting BAR0 and BAR1 both to
> > n, and saving the configuration, there is neither FOO0 nor FOO1 in
> > the resulting configuration file.
> 
> Indeed, that paragraph seems wrong. I think we should revert commit
> 64b81ed. Randy, Arnaud?

I just tested with Martin's example, and:
  - indeed, both symbols appear as "is not set",
  - that paragraph was already wrong for Kconfig at #64b81ed.

Before we trim that paragraph away, what's the purpose of the expected
difference in behavior for the two constructs?

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'