log -L/<regex>/,+1 not accepted (#1856)

Please see my original observation below.
Is it possible, to extend the git-log syntax in the way, that it accepts the short -L option (without :file) of blame in unique cases (only one file is logged or respectively the -L expression may be valid for all logged files)? It would be nice for command line users!

Alternatively I could also imagine the extension of the blame functionality in the direction to see a whole history instead of only the last modification.

Best regards,
Peter Dolland

-----------------------
Von: Johannes Schindelin [mailto:[email protected]]
Gesendet: Dienstag, 2. Oktober 2018 01:06
An: git-for-windows/git
Cc: Dolland Peter MTPCE; Mention
Betreff: Re: [git-for-windows/git] log -L/<regex>/,+1 not accepted (#1856)

It would be appropriate, but what @PhilipOakley asked you, @pdolland, was whether this is Windows-specific behavior. Because if it is not, then the feature request would be better sent to [email protected] (the real Git mailing list; please make sure to send plain-text-only mails). You saw fit to simply delete the issue reporting template instead of filling it out, so I have no way to assess whether WSL would be an option for you.

-----------
Von: Peter Dolland [mailto:[email protected]]
Gesendet: Montag, 1. Oktober 2018 13:26
An: git-for-windows/git
Cc: Dolland Peter MTPCE; Your activity
Betreff: Re: [git-for-windows/git] log -L/<regex>/,+1 not accepted (#1856)

I have no possibility to check this on Linux.

However, I found out, that the documentation of the –L option is different for blame and log. The documented log –L is working, even if the output is not exactly, what I wanted. Nevertheless, I think it would be appropriate, to admit the syntax of blame for log too.

Best regards,

Peter Dolland

Von: Philip Oakley [mailto:[email protected]]
Gesendet: Montag, 1. Oktober 2018 13:04
An: git-for-windows/git
Cc: Dolland Peter MTPCE; Author
Betreff: Re: [git-for-windows/git] log -L/<regex>/,+1 not accepted (#1856)

Have you been able to check if this issue is also present on Linux? It may be a a global Git problem rather than just a G-f-W issue.

I know there were some fairly recent upstream patches regarding the start,end aspects of the the '-L' option (IIRC mainly about the values spanning the end of the file).

There are a number of these options that do not carry fully between commands.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub<https://github.com/git-for-windows/git/issues/1856#issuecomment-425868832>, or mute the thread<https://github.com/notifications/unsubscribe-auth/ADShEBzWN4hO6uC4frkbHUWfjC0iOccyks5ugfa0gaJpZM4XBvep>.

--------- <the original> ------------
Von: Peter Dolland [mailto:[email protected]]
Gesendet: Montag, 1. Oktober 2018 12:01
An: git-for-windows/git
Cc: Dolland Peter MTPCE; Your activity
Betreff: [git-for-windows/git] log -L/<regex>/,+1 not accepted (#1856)

$ git log -L'/DRIVER_STATE = "/',+1 -- plm-dev/Wolke_M600_UTF8.java
fatal: -L argument not 'start,end:file' or ':funcname:file': /DRIVER_STATE = "/,+1
whereas blame with the same arguments does it:
$ git blame -L'/DRIVER_STATE = "/',+1 -- plm-dev/Wolke_M600_UTF8.java
51afb3c491 (Peter 2018-04-19 10:23:41 +0200 51) protected static final String DRIVER_STATE = "180419";
The same behavior with git versions 2.5.1 and 2.19.0.windows.1.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or mute the thread.

That would be nice, but I suspect in many cases the regex will be less unique than you might hope. E.g., if you're looking for the log of a particular function, you care about where it's defined. But unless you write your regex very carefully, you're going to also match places where it's called.

I have a hacky script (included below) that uses an already-built ctags index to pick the correct file.

Have you tried using a blame interface that supports parent-reblaming (i.e., once you blame a line to a particular commit, you can restart the blame from that commit's parent, digging further into history each time)? I use "tig blame" for this, and I find that I very rarely actually turn to "log -L".

-Peff

-- >8 --
#!/usr/bin/env perl

if (!@ARGV) {
print STDERR "usage: git flog [options] <function>\n"; exit 1;
}

my $func = pop @ARGV;
my $file = get_file_from_tags($func);
my $regex = '[^A-Za-z_]' . $func . '[^A-Za-z0-9_]'; exec qw(git log), "-L:$regex:$file", @ARGV; exit 1;

sub get_file_from_tags {
my $token = shift;

open(my $fh, '<', 'tags')

    or die "unable to open tags: $!\n";

while (<$fh>) {
chomp;

# this isn't exactly right, as the Ex command may contain
# embedded tabs, but it's accurate for the token and filename, # which come before, and probably good enough to match extension fields # which come after
my @fields = split /\t/;

next unless $fields[0] eq $token;

# only look for functions; assumes your ctags uses the "kind" # extension field. Note also that some implementations write the "kind:" # header and some do not. This handles both.
next unless grep { /^(kind:\s*)?f$/ } @fields;

# there may be more, but we don't have any way of disambiguating, # so just return the first match
return $fields[1];
}

die "unknown token: $token\n";
}