patchdoc/gitattributes: add Octave

> `matlab` pattern is also suitable for source code
> in the GNU Octave language.
>
> Signed-off-by: Boxuan Li <liboxuan@connect.hku.hk>
> ---
>   Documentation/gitattributes.txt | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/Documentation/gitattributes.txt b/Documentation/gitattributes.txt
> index 4fb20cd0e9..1b28381bda 100644
> --- a/Documentation/gitattributes.txt
> +++ b/Documentation/gitattributes.txt
> @@ -819,7 +819,7 @@ patterns are available:
>   
>   - `java` suitable for source code in the Java language.
>   
> -- `matlab` suitable for source code in the MATLAB language.
> +- `matlab` suitable for source code in the MATLAB/Octave language.
>   
>   - `objc` suitable for source code in the Objective-C language.
>   

> `matlab` pattern is also suitable for source code
> in the GNU Octave language.
>
> Signed-off-by: Boxuan Li <liboxuan@connect.hku.hk>
> ---
>  Documentation/gitattributes.txt | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/Documentation/gitattributes.txt b/Documentation/gitattributes.txt
> index 4fb20cd0e9..1b28381bda 100644
> --- a/Documentation/gitattributes.txt
> +++ b/Documentation/gitattributes.txt
> @@ -819,7 +819,7 @@ patterns are available:
>
>  - `java` suitable for source code in the Java language.
>
> -- `matlab` suitable for source code in the MATLAB language.
> +- `matlab` suitable for source code in the MATLAB/Octave language.
>
>  - `objc` suitable for source code in the Objective-C language.

>
>
> On Fri, May 10 2019, Boxuan Li wrote:
>
> > `matlab` pattern is also suitable for source code
> > in the GNU Octave language.
> >
> > Signed-off-by: Boxuan Li <liboxuan@connect.hku.hk>
> > ---
> >  Documentation/gitattributes.txt | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/Documentation/gitattributes.txt b/Documentation/gitattributes.txt
> > index 4fb20cd0e9..1b28381bda 100644
> > --- a/Documentation/gitattributes.txt
> > +++ b/Documentation/gitattributes.txt
> > @@ -819,7 +819,7 @@ patterns are available:
> >
> >  - `java` suitable for source code in the Java language.
> >
> > -- `matlab` suitable for source code in the MATLAB language.
> > +- `matlab` suitable for source code in the MATLAB/Octave language.
> >
> >  - `objc` suitable for source code in the Objective-C language.
>
> I've never used either, but browsing our patterns I see:
>
>          "^[[:space:]]*((classdef|function)[[:space:]].*)$|^%%[[:space:]].*$",
>
> I.e. that last bit matches ^%%, and then here:
> https://en.wikibooks.org/wiki/MATLAB_Programming/Differences_between_Octave_and_MATLAB
>
>     MATLAB uses the percent sign '%' to begin a comment. Octave uses
>     both the hash symbol # and the percent sign % interchangeably.
>
> So here we have the "function" pattern matching a comment, right? (this
> doesn't have any tests) and we'd want to add "#" to Octave, but not
> MATLAB.

> Signed-off-by: Boxuan Li <liboxuan@connect.hku.hk>
> ---
> Update from v2: fix indentation
> ---
>  Documentation/gitattributes.txt | 2 ++
>  userdiff.c                      | 5 +++++
>  2 files changed, 7 insertions(+)
>
> diff --git a/Documentation/gitattributes.txt b/Documentation/gitattributes.txt
> index 4fb20cd0e9..45374c7dd3 100644
> --- a/Documentation/gitattributes.txt
> +++ b/Documentation/gitattributes.txt
> @@ -821,6 +821,8 @@ patterns are available:
>  
>  - `matlab` suitable for source code in the MATLAB language.
>  
> +- `octave` suitable for source code in the Octave language.
> +
>  - `objc` suitable for source code in the Objective-C language.
>  
>  - `pascal` suitable for source code in the Pascal/Delphi language.
> diff --git a/userdiff.c b/userdiff.c
> index 3a78fbf504..7d07b82116 100644
> --- a/userdiff.c
> +++ b/userdiff.c
> @@ -60,6 +60,11 @@ PATTERNS("java",
>  PATTERNS("matlab",
>  	 "^[[:space:]]*((classdef|function)[[:space:]].*)$|^%%[[:space:]].*$",
>  	 "[a-zA-Z_][a-zA-Z0-9_]*|[-+0-9.e]+|[=~<>]=|\\.[*/\\^']|\\|\\||&&"),
> +PATTERNS("octave",
> +	 /* Mostly the same as matlab. In addition, Octave
> +	  * supports '##' and '%%%' for code sections */
> +	 "^[[:space:]]*((classdef|function)[[:space:]].*)$|^(%%%?|##)[[:space:]].*$",
> +	 "[a-zA-Z_][a-zA-Z0-9_]*|[-+0-9.e]+|[=~<>]=|\\.[*/\\^']|\\|\\||&&"),
>  PATTERNS("objc",
>  	 /* Negate C statements that can look like functions */
>  	 "!^[ \t]*(do|for|if|else|return|switch|while)\n"

> @@ -60,6 +60,11 @@ PATTERNS("java",
>  PATTERNS("matlab",
>  	 "^[[:space:]]*((classdef|function)[[:space:]].*)$|^%%[[:space:]].*$",
>  	 "[a-zA-Z_][a-zA-Z0-9_]*|[-+0-9.e]+|[=~<>]=|\\.[*/\\^']|\\|\\||&&"),
> +PATTERNS("octave",
> +	 /* Mostly the same as matlab. In addition, Octave
> +	  * supports '##' and '%%%' for code sections */
> +	 "^[[:space:]]*((classdef|function)[[:space:]].*)$|^(%%%?|##)[[:space:]].*$",
> +	 "[a-zA-Z_][a-zA-Z0-9_]*|[-+0-9.e]+|[=~<>]=|\\.[*/\\^']|\\|\\||&&"),
>  PATTERNS("objc",
>  	 /* Negate C statements that can look like functions */
>  	 "!^[ \t]*(do|for|if|else|return|switch|while)\n"
> 

>
> Am 11.05.19 um 06:13 schrieb Boxuan Li:
> > Octave pattern is almost the same as matlab. Besides,
> > octave also uses '%%%' or '##' to begin code sections.
> >
>
> > @@ -60,6 +60,11 @@ PATTERNS("java",
> >  PATTERNS("matlab",
> >        "^[[:space:]]*((classdef|function)[[:space:]].*)$|^%%[[:space:]].*$",
> >        "[a-zA-Z_][a-zA-Z0-9_]*|[-+0-9.e]+|[=~<>]=|\\.[*/\\^']|\\|\\||&&"),
> > +PATTERNS("octave",
> > +      /* Mostly the same as matlab. In addition, Octave
> > +       * supports '##' and '%%%' for code sections */
> > +      "^[[:space:]]*((classdef|function)[[:space:]].*)$|^(%%%?|##)[[:space:]].*$",
> > +      "[a-zA-Z_][a-zA-Z0-9_]*|[-+0-9.e]+|[=~<>]=|\\.[*/\\^']|\\|\\||&&"),
> >  PATTERNS("objc",
> >        /* Negate C statements that can look like functions */
> >        "!^[ \t]*(do|for|if|else|return|switch|while)\n"
> >
>
> In Matlab, are %%% and ## valid syntax? If not, instead of introducing a
> new language, please just extend the Matlab rule to treat %%% and ## as
> you need for Octave and mark your Octave files as Matlab.
>
> -- Hannes

> On Wed, May 15, 2019 at 1:57 PM Johannes Sixt <j6t@kdbg.org> wrote:
>>
>> Am 11.05.19 um 06:13 schrieb Boxuan Li:
>>> Octave pattern is almost the same as matlab. Besides,
>>> octave also uses '%%%' or '##' to begin code sections.
>>>
>>
>>> @@ -60,6 +60,11 @@ PATTERNS("java",
>>>  PATTERNS("matlab",
>>>        "^[[:space:]]*((classdef|function)[[:space:]].*)$|^%%[[:space:]].*$",
>>>        "[a-zA-Z_][a-zA-Z0-9_]*|[-+0-9.e]+|[=~<>]=|\\.[*/\\^']|\\|\\||&&"),
>>> +PATTERNS("octave",
>>> +      /* Mostly the same as matlab. In addition, Octave
>>> +       * supports '##' and '%%%' for code sections */
>>> +      "^[[:space:]]*((classdef|function)[[:space:]].*)$|^(%%%?|##)[[:space:]].*$",
>>> +      "[a-zA-Z_][a-zA-Z0-9_]*|[-+0-9.e]+|[=~<>]=|\\.[*/\\^']|\\|\\||&&"),
>>>  PATTERNS("objc",
>>>        /* Negate C statements that can look like functions */
>>>        "!^[ \t]*(do|for|if|else|return|switch|while)\n"
>>>
>>
>> In Matlab, are %%% and ## valid syntax? If not, instead of introducing a
>> new language, please just extend the Matlab rule to treat %%% and ## as
>> you need for Octave and mark your Octave files as Matlab.
> 
> '##' is not valid syntax in Matlab scripts.
> 
> '%%%' is valid syntax in Matlab. However, it is not used as a section divider.

> Johannes Sixt <j6t@kdbg.org> writes:
> 
>> In Matlab, is %%% followed by space at the beginning of a line
>> *commonly* used for something different? If I were to make a guess, I
>> would say no. If I'm right, it does not hurt to merge the Octave rules
>> into the Matlab rules.
> 
> That is true because we are not syntax-aware and error-highlighting
> text editor.  If we were, I'd suspect that your stance may probably
> be different.  But instead we apply these patterns to a program that
> is assumed to be correctly written.

> And from that point of view, I agree with you that it would not hurt
> to make the existing patterns for Matlab slightly more receptive so
> that a correctly written programs in either language would be matched
> appropriately.
> 
> But would it hurt to have two similar entries, with a clear
> description on how they are different, in our code there, given how
> infrequently individual entries have historically been updated?

>
> Johannes Sixt <j6t@kdbg.org> writes:
>
> > I'd prefer to keep this list at the minimum necessary as long as it is
> > hard-coded in C.
>
> Yeah, I know that feeling.
>

> > I would take a different stance if this were some
> > configuration file that we ship.
>
> Hmm, now you reminded me of my ancient wish.
>
> Perhaps it is not too bad to ship $(sharedir)/git-core/userdiff that
> can be read using git_config_from_file() interface, using a very
> narrow callback function that understands only diff.*.xfuncname and
> diff.*.wordregex and discards everything else, without even
> following the include/includeIf stuff?  As long as that can be done
> safely and without too much overhead, we could get rid of the
> hardcoded patterns in userdiff.c::builtin_drivers[] and that would
> be wonderful ;-)
>

> On Fri, May 17, 2019 at 7:34 AM Junio C Hamano <gitster@pobox.com> wrote:
>>
>> Johannes Sixt <j6t@kdbg.org> writes:
>>
>>> I'd prefer to keep this list at the minimum necessary as long as it is
>>> hard-coded in C.
>>
>> Yeah, I know that feeling.
> 
> So do we reach a consensus? Should I merge the Octave rules into Matlab rules?

> Octave pattern is almost the same as matlab, except
> that '%%%' and '##' can also be used to begin code sections,
> in addition to '%%' that is understood by both. Octave
> pattern is merged into Matlab pattern. Test cases for
> the hunk header patterns of matlab and octave under
> t/t4018 are added.

> 
> Signed-off-by: Boxuan Li <liboxuan@connect.hku.hk>
> ---
> v1: use matlab pattern for octave
> v2: add a new octave pattern
> v3: fix indentation problem of v2
> v4: improve doc and commit message
> v5: merge octave pattern into matlab and add test cases
> ---

> diff --git a/Documentation/gitattributes.txt b/Documentation/gitattributes.txt
> index 4fb20cd0e9..e657a6b7b7 100644
> --- a/Documentation/gitattributes.txt
> +++ b/Documentation/gitattributes.txt
> @@ -819,7 +819,7 @@ patterns are available:
>  
>  - `java` suitable for source code in the Java language.
>  
> -- `matlab` suitable for source code in the MATLAB language.
> +- `matlab` suitable for source code in the MATLAB and Octave language.

> diff --git a/userdiff.c b/userdiff.c
> index 3a78fbf504..bcae006abe 100644
> --- a/userdiff.c
> +++ b/userdiff.c
> @@ -58,7 +58,10 @@ PATTERNS("java",
>  	 "|[-+*/<>%&^|=!]="
>  	 "|--|\\+\\+|<<=?|>>>?=?|&&|\\|\\|"),
>  PATTERNS("matlab",
> -	 "^[[:space:]]*((classdef|function)[[:space:]].*)$|^%%[[:space:]].*$",
> +	 /* Octave pattern is mostly the same as matlab, except that '%%%' and
> +	  * '##' can also be used to begin code sections, in addition to '%%'
> +	  * that is understood by both. */

> +	 "^[[:space:]]*((classdef|function)[[:space:]].*)$|^(%%%?|##)[[:space:]].*$",
>  	 "[a-zA-Z_][a-zA-Z0-9_]*|[-+0-9.e]+|[=~<>]=|\\.[*/\\^']|\\|\\||&&"),
>  PATTERNS("objc",
>  	 /* Negate C statements that can look like functions */
> 

> Johannes Sixt <j6t@kdbg.org> writes:
> 
>> Junio will very likely appreciate if you resend with these fixes applied.
> 
> Heh, thanks, for spotting, but this round has already been in 'next'
> so any updates should be done incrementally X-<.
>