Re: [PATCH] Add --pretty=changelog

From: Simon Josefsson <simon@josefsson.org>
Date: Mar 1, 2007, 15:23 UTC
Message-ID: <87y7mhrnrc.fsf@latte.josefsson.org>
In-Reply-To: <Pine.LNX.4.63.0702271621120.22628@wbgn013.biozentrum.uni-wuerzburg.de>

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

> This outputs the log in GNU ChangeLog format.

Many thanks! I ran this in my cvs->git repository for libtasn1, and then did a diff against the output from cvs2cl in the old repository. Here are some categories of differences:

1) Indentation should be under the '*'.  Compare (wrong):

        * src/asn1Coding.c, src/asn1Decoding.c, src/asn1Parser.c: Improve
          --help output.  Assume getopt_long, since gnulib provides it.

with (correct):

        * src/asn1Coding.c, src/asn1Decoding.c, src/asn1Parser.c: Improve
        --help output.  Assume getopt_long, since gnulib provides it.

2) Don't log anything for empty messages:

        * gl/.cvsignore: *** empty log message ***

Possibly, this should be an option.

3) Possible charset problem?  Compare this (correct):

        * configure.in: Fix -Wno-pointer-sign test to respect user-defined
        CFLAGS.  Reported by "Diego 'Flameeyes' Pettenò"
        <flameeyes@gentoo.org>.

with (wrong):

        * configure.in: Fix -Wno-pointer-sign test to respect user-defined
          CFLAGS.  Reported

The git log do seem to contain the correct data, though, from git-log:

    Fix -Wno-pointer-sign test to respect user-defined CFLAGS.  Reported
    by "Diego 'Flameeyes' Pettenò" <flameeyes@gentoo.org>.

4) Weird error, compare (correct):

        * lib/coding.c:
        (asn1_der_coding): For TYPE_NULL, increment counter even if we don't
        write any output.  Reported by Stephen Wrobleski
        <steve@localtoast.org>.

with (wrong):

        * lib/coding.c: (asn1_der_coding): For TYPE_NULL, increment counter
          even if we don't

Maybe this is related to the last one, and the problem really is '<' or '>'?

Here is another one, compare (correct):

        * lib/coding.c:
        (asn1_octet_der): Work even if str_len is 0, i.e., write an ASN.1
        length of zero.  Otherwise encodings became garbled on 64-bit
        platforms, detected while running the Shishi self-tests on the
        Debian build robots.  A self test to reproduce this is in Test_tree.

with (wrong):

        * lib/coding.c: (asn1_octet_der): Work even if str_len is 0, i.e.,
          write an ASN.1

I'll let you fix these and resume the diff later. :)

Can I pull your git tree from somewhere? Applying patches from e-mail is kind of boring.

/Simon