[BUG?] blame: Odd -L 1,+0 behavior

[BUG?] blame: Odd -L 1,+0 behavior

[Ævar Arnfjörð Bjarmason]

On Fri, Jul 16, 2010 at 15:35, Ævar Arnfjörð Bjarmason <avarab@gmail.com> wrote:
> git-pickaxe (later git-blame) gained support for the -L/start/,/end/
> form in 2006 (931233bc66 by Junio C Hamano), but nothing was added to
> test this functionality. Change that by adding more -L tests to
> t8003-blame.sh.

I was also going to patch t8003-blame.sh to test for the functionality
introduced in 7bd9641df5, but I'm not sure what it's supposed to do:

ok:

    $ git blame cow
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 1) ABC
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 2) DEF
    018e3134 (Fifth  2005-04-07 15:17:13 -0700 3) XXXX
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 4) GHIJK

ok:

    $ git blame -L 1,2 cow
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 1) ABC
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 2) DEF

ok, +/- are zero-indexed:

    $ git blame -L 1,+2 cow
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 1) ABC
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 2) DEF

Shouldn't this either print nothing, er be an error:

    $ git blame -L 1,+0 cow
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 1) ABC
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 2) DEF

same here, line 0-0 prints the whole file:

    $ git blame -L 0,0 cow
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 1) ABC
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 2) DEF
    018e3134 (Fifth  2005-04-07 15:17:13 -0700 3) XXXX
    5b22ca36 (Fourth 2005-04-07 15:16:13 -0700 4) GHIJK

And these need a better error message:

    $ git blame -L 0,0a cow
    usage: git blame [options] [rev-opts] [rev] [--] file

Is there any use for +/- 0, or should it always be an error?

Re: [BUG?] blame: Odd -L 1,+0 behavior

[Junio C Hamano]

Ævar Arnfjörð Bjarmason <avarab@gmail.com> writes:

> ok, +/- are zero-indexed:
>
>     $ git blame -L 1,+2 cow

I don't know what "zero-indexed" means, but +2 means "starting from the
line I told you earlier, give me two lines".  Likewise -2 means "ending at
the line I told you earlier, give me two lines".  As a side effect of the
internal implementation of this logic -L 5,3 means the same thing as -L 3,5
but that is not an intended nor documented behaviour.

> Shouldn't this either print nothing, er be an error:
(multiple)

The parsing code is lax in the sense that rejecting nonsensical input like
"-L 10,-100" and "-L 2,+0" as an error was not considered a primary goal.
The only error checking it does is to make sure it does not parse numbers
that it cannot use (i.e. start from line 30 in a file that does not have
that many lines).

Re: [BUG?] blame: Odd -L 1,+0 behavior

[Ævar Arnfjörð Bjarmason]

On Fri, Jul 16, 2010 at 21:11, Junio C Hamano <gitster@pobox.com> wrote:
> Ævar Arnfjörð Bjarmason <avarab@gmail.com> writes:
>
>> ok, +/- are zero-indexed:
>>
>>     $ git blame -L 1,+2 cow
>
> I don't know what "zero-indexed" means, but +2 means "starting from the
> line I told you earlier, give me two lines".  Likewise -2 means "ending at
> the line I told you earlier, give me two lines".

From just using it I parsed 1,+1 as "give me line #1 plus $n-1
lines". At first I thought 1,+1 would print two lines (line 1 *plus*
one). But it makes sense when you explain it.

> As a side effect of the internal implementation of this logic -L 5,3
> means the same thing as -L 3,5 but that is not an intended nor
> documented behaviour.

How should L 5,3 and 3,5 work? Should the former give an error?

>> Shouldn't this either print nothing, er be an error:
> (multiple)
>
> The parsing code is lax in the sense that rejecting nonsensical input like
> "-L 10,-100" and "-L 2,+0" as an error was not considered a primary goal.
> The only error checking it does is to make sure it does not parse numbers
> that it cannot use (i.e. start from line 30 in a file that does not have
> that many lines).

Do you want a patch to make it less lax?

Re: [BUG?] blame: Odd -L 1,+0 behavior

[Junio C Hamano]

Ævar Arnfjörð Bjarmason <avarab@gmail.com> writes:

>> As a side effect of the internal implementation of this logic -L 5,3
>> means the same thing as -L 3,5 but that is not an intended nor
>> documented behaviour.
>
> How should L 5,3 and 3,5 work? Should the former give an error?

"-L 3,5" would choose "lines 3,4, and 5" (both ends inclusive).  Currently
"-L 5,3" does the same thing but as I said, that is not an intended nor
documented behaviour, and if you are tackling this area to tighten the
parsing and error diagnosis, I think it is reasonable to error it out.

>>> Shouldn't this either print nothing, er be an error:
>> (multiple)
>>
>> The parsing code is lax in the sense that rejecting nonsensical input like
>> "-L 10,-100" and "-L 2,+0" as an error was not considered a primary goal.
>> The only error checking it does is to make sure it does not parse numbers
>> that it cannot use (i.e. start from line 30 in a file that does not have
>> that many lines).
>
> Do you want a patch to make it less lax?

Be my guest ;-)  Thanks.