patch, 2 partsMinor userdiff stuff

> All diffs following a function definition will have that function name
> as chunck header, but this is the best we can do with the current
> userdiff capabilities.
>
> Signed-off-by: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
> ---
>  userdiff.c |    3 +++
>  1 files changed, 3 insertions(+), 0 deletions(-)
>
> diff --git a/userdiff.c b/userdiff.c
> index 01d3a8b..70120c3 100644
> --- a/userdiff.c
> +++ b/userdiff.c
> @@ -107,6 +107,9 @@
>  	 "(@|@@|\\$)?[a-zA-Z_][a-zA-Z0-9_]*"
>  	 "|[-+0-9.e]+|0[xXbB]?[0-9a-fA-F]+|\\?(\\\\C-)?(\\\\M-)?."
>  	 "|//=?|[-+*/<>%&^|=!]=|<<=?|>>=?|===|\\.{1,3}|::|[!=]~"),
> +PATTERNS("shell", "^[ \t]*([a-zA-Z_0-9]+)[ \t]*\\(\\).*",
> +	 /* -- */
> +	 "(--|\\$)?[a-zA-Z_0-9]+|&&|\\|\\|"),

> Giuseppe Bilotta <giuseppe.bilotta@gmail.com> writes:
>
>> All diffs following a function definition will have that function name
>> as chunck header, but this is the best we can do with the current
>> userdiff capabilities.
>>
>> Signed-off-by: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
>> ---
>>  userdiff.c |    3 +++
>>  1 files changed, 3 insertions(+), 0 deletions(-)
>>
>> diff --git a/userdiff.c b/userdiff.c
>> index 01d3a8b..70120c3 100644
>> --- a/userdiff.c
>> +++ b/userdiff.c
>> @@ -107,6 +107,9 @@
>>        "(@|@@|\\$)?[a-zA-Z_][a-zA-Z0-9_]*"
>>        "|[-+0-9.e]+|0[xXbB]?[0-9a-fA-F]+|\\?(\\\\C-)?(\\\\M-)?."
>>        "|//=?|[-+*/<>%&^|=!]=|<<=?|>>=?|===|\\.{1,3}|::|[!=]~"),
>> +PATTERNS("shell", "^[ \t]*([a-zA-Z_0-9]+)[ \t]*\\(\\).*",
>> +      /* -- */
>> +      "(--|\\$)?[a-zA-Z_0-9]+|&&|\\|\\|"),
>
> Hmm, what is this "double-dash -- might be present before a name" about?
>

> On Wed, Aug 03, 2011 at 07:26:16AM +0200, Giuseppe Bilotta wrote:
>
>> All diffs following a function definition will have that function name
>> as chunck header, but this is the best we can do with the current
>> userdiff capabilities.
>
> Curious as to how this would look in git.git, I tried "git log -p"
> before and after your patches, and diffed the result. I noticed two
> things:
>
>  1. Given a block of shell code like this:
>
>        foo() {
>          ... do something ...
>        }
>
>        test_expect_success 'test foo' '
>          ... the actual test ...
>        '
>
>     if we add new code after the test, the old regex would print:
>
>        @@ -1,2 +3,4 @@ test_expect_success 'test foo' '
>
>     and now we say:
>
>        @@ -1,2 +3,4 @@ foo
>
>     which seems more misleading. I know the function-matching code has
>     no way to say "look for ^}, which signals end of function", so we
>     can't be entirely accurate. But I wonder if the new heuristic
>     (which seems to look for a name followed by parentheses) is
>     actually any better than the old.

>  2. What would have printed before:
>
>       @@ -1,2 +3,4 @@ foo() {
>
>     now prints
>
>       @@ -1,2 +3,4 @@ foo
>
>     without the parentheses or brace. It looks like the similar C one
>     keeps the parentheses, at least. I find that a bit more readable,
>     as it is more clear that the line indicates a function, and not
>     simply some top-level command.

> On Wed, Aug 03, 2011 at 07:26:16AM +0200, Giuseppe Bilotta wrote:
>
>> All diffs following a function definition will have that function name
>> as chunck header, but this is the best we can do with the current
>> userdiff capabilities.
>
> Curious as to how this would look in git.git, I tried "git log -p"
> before and after your patches, and diffed the result. I noticed two
> things: