git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH] cvsimport: rewrite to use cvsps 3.x to fix major bugs

From
Junio C Hamano <gitster@pobox.com>
Date
Jan 11, 2013, 16:31 UTC
Message-ID
<7va9sfd6lk.fsf@alter.siamese.dyndns.org>
In-Reply-To
<1357875152-19899-1-git-send-email-gitster@pobox.com>
Show 25 quoted lines
> From: "Eric S. Raymond" <esr@thyrsus.com>
> ...
> diff --git a/git-cvsimport.perl b/git-cvsimport-fallback.perl
> similarity index 98%
> rename from git-cvsimport.perl
> rename to git-cvsimport-fallback.perl
> index 0a31ebd..4bc0717 100755
> --- a/git-cvsimport.perl
> +++ b/git-cvsimport-fallback.perl
> @@ -1,4 +1,8 @@
>  #!/usr/bin/perl
> +# This code became obsolete in January 2013, and is retained only as a
> +# fallback from git-cvsimport.py for users who have only cvsps-2.x.
> +# It (and the code in cvsimport.py that calls it) should be removed
> +# once the 3.x version has had a reasonable time to propagate.
>  
>  # This tool is copyright (c) 2005, Matthias Urlichs.
>  # It is released under the Gnu Public License, version 2.
> @@ -27,6 +31,10 @@
>  use POSIX qw(strftime tzset dup2 ENOENT);
>  use IPC::Open2;
>  
> +print(STDERR "You do not appear to have cvsps 3.x.\n");
> +print(STDERR "Falling back to unmaintained Perl cvsimport for cvsps 2.x.\n");
> +print(STDERR "Upgrade your cvsps for best results.\n");

I think the prevalent style in this script is to write "print" without parentheses:

	print STDERR "msg\n";
Show 24 quoted lines
> diff --git a/git-cvsimport.py b/git-cvsimport.py
> new file mode 100755
> index 0000000..129471e
> --- /dev/null
> +++ b/git-cvsimport.py
> @@ -0,0 +1,354 @@
> +#!/usr/bin/env python
> +#
> +# Import CVS history into git
> +#
> +# Intended to be a near-workalike of Matthias Urlichs's Perl implementation.
> ...
> +class cvsps:
> +    "Method class for cvsps back end."
> +    def __init__(self):
> +        self.opts = ""
> +        self.revmap = None
> +    def set_repo(self, val):
> +        "Set the repository root option."
> +        if not val.startswith(":"):
> +            if not val.startswith(os.sep):
> +                val = os.path.abspath(val)
> +            val = ":local:" + val
> +        self.opts += " --root '%s'" % val

This looks lazy and unsafe quoting. Is there anything that makes sure repository path does not contain a single quote?

Show 12 quoted lines
> +    def set_authormap(self, val):
> +        "Set the author-map file."
> +        self.opts += " -A '%s'" % val
> +    def set_fuzz(self, val):
> +        "Set the commit-similarity window."
> +        self.opts += " -z %s" % val
> +    def set_nokeywords(self):
> +        "Suppress CVS keyword expansion."
> +        self.opts += " -k"
> +    def add_opts(self, val):
> +        "Add options to the engine command line."
> +        self.opts += " " + val
... especially for callers of this method.

The same comment applies to many uses of "val" in the method implementations of this class and the cvs2git class.

> +    def command(self):
> +        "Emit the command implied by all previous options."
> +        return "(cvs2git --username=git-cvsimport --quiet --quiet --blobfile={0} --dumpfile={1} {2} {3} && cat {0} {1} && rm {0} {1})".format(tempfile.mkstemp()[1], tempfile.mkstemp()[1], self.opts, self.modulepath)
Could we do something better with this overlong source line?
Show 24 quoted lines
> +if __name__ == '__main__':
> ...
> +    for (opt, val) in options:
> +        if opt == '-v':
> +            verbose += 1
> +        elif opt == '-b':
> +            bare = True
> +        elif opt == '-e':
> +            for cls in (cvsps, cvs2git):
> +                if cls.__name__ == val:
> +                    backend = cls()
> +                    break
> +            else:
> +                sys.stderr.write("git cvsimport: unknown engine %s.\n" % val)
> +                sys.exit(1)
> +        elif opt == '-d':
> +            backend.set_repo(val)
> +        elif opt == '-C':
> +            outdir = val
> +        elif opt == '-r':
> +            remotize = True
> +        elif opt == '-o':
> +            sys.stderr.write("git cvsimport: -o is no longer supported.\n")
> +            sys.exit(1)
Isn't this a regression?
Show 15 quoted lines
> +        elif opt == '-i':
> +            import_only = True
> +        elif opt == '-k':
> +            backend.set_nokeywords()
> +        elif opt == '-u':
> +            underscore_to_dot = True
> +        elif opt == '-s':
> +            slashsubst = val
> +        elif opt == '-p':
> +            backend.add_opts(val.replace(",", " "))
> +        elif opt == '-z':
> ...
> +        elif opt == '-P':
> +            backend = filesource(val)
> +            sys.exit(1)
???
> +        elif opt in ('-m', '-M'):
> +            sys.stderr.write("git cvsimport: -m and -M are no longer supported: use reposurgeon instead.\n")
> +            sys.exit(1)

I wonder if it is better to ignore these options with a warning but still let the command continue; cvsps-3.x was supposed to get merges right without the help of these ad-hoc options, no?

Otherwise it looks like a regression to me.
Show 36 quoted lines
> +        elif opt == '-S':
> +            backend.set_exclusion(val)
> +        elif opt == '-a':
> +            sys.stderr.write("git cvsimport: -a is no longer supported.\n")
> +            sys.exit(1)
> +        elif opt == '-L':
> +            sys.stderr.write("git cvsimport: -L is no longer supported.\n")
> +            sys.exit(1)
> +        elif opt == '-A':
> +            authormap = os.path.abspath(val)
> +        elif opt == '-R':
> +            revisionmap = True
> +        else:
> +            print """\
> +git cvsimport [-A <author-conv-file>] [-C <git_repository>] [-b] [-d <CVSROOT>]
> +     [-e engine] [-h] [-i] [-k] [-p <options-for-cvsps>] [-P <source-file>]
> +     [-r <remote>] [-R] [-s <subst>] [-S <regex>] [-u] [-v] [-z <fuzz>]
> +     [<CVS_module>]
> +"""
> +    def metadata(fn, outdir='.'):
> +        if bare:
> +            return os.path.join(outdir, fn)
> +        else:
> +            return os.path.join(outdir, ".git", fn)
> +    # Ugly fallback code for people with only cvsps-2.x
> +    # Added January 2013 - should be removed after a decent interval.
> +    if backend.__class__.__name__ == "cvsps":
> +        try:
> +            subprocess.check_output("cvsps -V 2>/dev/null", shell=True)
> +        except subprocess.CalledProcessError as e:
> +            if e.returncode == 1:
> +                sys.stderr.write("cvsimport: falling back to old version...\n")
> +                sys.exit(os.system("git-cvsimport-fallback " + " ".join(sys.argv[1:])))
> +            else:
> +                sys.stderr.write("cvsimport: cannot execute cvsps.\n")
> +                sys.exit(1)

Having the code to die when it sees options the rewritten version does not yet support before it calls the fallback makes the fallback much less effective, no?

Show 8 quoted lines
> +    # Real mainline code begins here
> +    try:
> +        if outdir:
> +            try:
> +                # If the output directory does not exist, create it
> +                # and initialize it as a git repository.
> +                os.mkdir(outdir)
> +                do_or_die("git init --quiet " + outdir)

Did anything made sure outdir is without $IFS chars up to this point?

Not very impressed (yet). The advertised "fix major bugs" sounds more like "trade major bugs with different ones with a couple of feature removals" at this point.

Previous: Junio C HamanoNext: Eric S. Raymond
Message 2 of 37 in “cvsimport: rewrite to use cvsps 3.x to fix major bugs”
  1. cvsimport: rewrite to use cvsps 3.x to fix major bugsJunio C Hamano, Jan 11, 2013
  2. Junio C HamanoJan 11, 2013
  3. Eric S. RaymondJan 11, 2013
  4. Junio C HamanoJan 11, 2013
  5. Junio C HamanoJan 11, 2013
  6. Eric S. RaymondJan 12, 2013
  7. Junio C HamanoJan 12, 2013
  8. t/t960[123]: remove leftover scriptsJunio C Hamano, Jan 12, 2013
  9. Chris RorvickJan 12, 2013
  10. t/lib-cvs: cvsimport no longer works without Python >= 2.7Junio C Hamano, Jan 12, 2013
  11. Michael HaggertyJan 12, 2013
  12. John KeepingJan 13, 2013
  13. Eric S. RaymondJan 12, 2013
  14. Michael HaggertyJan 12, 2013
  15. Eric S. RaymondJan 12, 2013
  16. Jonathan NiederJan 12, 2013
  17. Jonathan NiederJan 12, 2013
  18. Junio C HamanoJan 13, 2013
  19. Junio C HamanoJan 13, 2013
  20. 0/3 A smoother transition plan for cvsimportJunio C Hamano, Jan 14, 2013
  21. 1/3 cvsimport: allow setting a custom cvsps (2.x) program nameJunio C Hamano, Jan 14, 2013
  22. 2/3 cvsimport: introduce a version-switch wrapperJunio C Hamano, Jan 14, 2013
  23. Junio C HamanoJan 14, 2013
  24. 3/3 cvsimport: start adding cvsps 3.x supportJunio C Hamano, Jan 14, 2013
  25. Chris RorvickJan 15, 2013
  26. Junio C HamanoJan 15, 2013
  27. Michael HaggertyJan 14, 2013
  28. 0/6 A smoother transition plan for cvsimportJunio C Hamano, Jan 14, 2013
  29. 1/6 Makefile: add description on PERL/PYTHON_PATHJunio C Hamano, Jan 14, 2013
  30. 2/6 cvsimport: allow setting a custom cvsps (2.x) program nameJunio C Hamano, Jan 14, 2013
  31. 3/6 cvsimport: introduce a version-switch wrapperJunio C Hamano, Jan 14, 2013
  32. 4/6 cvsimport: start adding cvsps 3.x supportJunio C Hamano, Jan 14, 2013
  33. 5/6 cvsimport: make tests reusable for cvsimport-3Junio C Hamano, Jan 14, 2013
  34. 6/6 cvsimport-3: add a sample testJunio C Hamano, Jan 14, 2013
  35. Jonathan NiederJan 14, 2013
  36. 7/6 t9600: further prepare for sharingJunio C Hamano, Jan 14, 2013
  37. 8/6 t9600: adjust for new cvsimportJunio C Hamano, Jan 14, 2013

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.