Re: [lmi] Nomenclature for the new md5sum

[Vadim Zeitlin]

On Mon, 17 Jun 2019 15:59:45 +0000 Greg Chicares <address@hidden> wrote:

GC> I guess the safest answer is to call ours 'lmi_md5sum.exe' and place
GC> it in 'third_party/bin/'.

 OK, we'll do this.

GC> Then it's just a drop-in replacement for the
GC> binary that we have always obtained from some external URL.

 It's not quite a "drop-in replacement" if it uses a different name. E.g.
if there are any existing scripts using the files from the fardel, they
would break. But I don't really know if any such scripts exist, of course.

GC> That way, nothing outside that very narrow use case can go wrong; and
GC> if anything does go wrong in that case, it's trivial to revert to the
GC> old binary.
GC> 
GC> At some later date, we could reconsider whatever cygwin or wine issues
GC> might arise, and perhaps decide to call it something else and install it
GC> somewhere else. But that requires deeper thought to foresee any problems
GC> that might arise. Time slots for deep thought are scarce.

 One issue which is already known is that we'll need to modify the unit
tests using md5sum external executable, i.e. system_command and
authenticity ones, to use lmi_md5sum instead, as otherwise they wouldn't
work under Wine (under native MSW or in a native Linux build we could just
use the existing standard md5sum binary, but trying to execute it from a
MSW executable running under Wine on Linux would fail). But this is just
for your information and doesn't change anything, of course.

 Regards,
VZ

pgpDwMqDaanSy.pgp
Description: PGP signature