On the "makepatch" target (Re: svn commit: r384004 - head/security/pecl-crack/files)

[Mikhail T.]

On 14.04.2015 13:09, Alexey Dokuchaev wrote:
> The point is to avoid useless noise in the diff.  "Statistics" you're
> talking about is pretty meaningless indeed, and since each next careless
> submitter will probably overwrite it with their own timezone makes it
> even more meaningless.
The next careless submitter would have to be patching the same file -- which
does not happen very often. But I understand your point as far the as timestamp
on the /original/ file goes.

I just tried using the target and found, that I dislike it for the following
reasons:

  * It retains the .orig suffix in the diff -- a completely useless 5 bytes
    carrying no information;
  * It discards the timestamp of the new version of the patched file, throwing
    out not only the timezone, but the actual time the work was done;
  * It creates, mechanically, one patch per file -- whereas I, for one, often
    prefer to group related changes to multiple files into a single patch (makes
    it easier to refer upstream maintainers to the patches);
  * The new patchs' names retain extensions of the patched files, such as, for
    example patch-crack.c. This confuses various tools, which treat files based
    on extensions like cscope or mantis bug-tracker, which, in this example,
    treat it like C-source, rather than a patch.

Yours,

    -mi