Re: Could timestamps be replaced with MD5?

[Brian Huddleston]

> As I already wrote in a previous posting I think it is really a good idea.
> Right now I have seen a problem in it: CVS Windows clients automatically
> convert text file line endings from LF to CRLF and vice versa. So
checksums
> computed on a Windows client will bedifferent from checksums computed on
the
> (*nix) server.

I don't think this is necessarily a fatal problem.  I've got md5sum on NT
and Linux
and the same build.xml file generates the same checksum.  (Despite LF on
Linux and
CRLF on NT).  So at least someone else has kludged around this problem.

For the purpose that it's being used, just canonicalizing CRLF to LF for
purposes of the
bytes fed to the digest would produce wrong (arguably) but consistent (the
important part) results.

I'm a little confused though, does the MD5 need to be the same server and
client side?  I thought
it would just be just be in the Entries file.  Perhaps I'm missing part of
the picture?  (Forgive me if this has been discussed.  I only skimmed the
thread.)

Brian Huddleston
Huddleston Consulting

>
> Best regards
> Martin
>
>