Re: [lmi] Bad md5sum.exe link

[Greg Chicares]

On 2019-04-29 21:55, Vadim Zeitlin wrote:
> 
>  One minor problem found during the testing of lmi installation (I'll write
> another email about the major problem soon) is that the link for md5sum.exe
> used in install_miscellanea.make is dead and trying to fetch it from
> http://etree.org/cgi-bin/counter.cgi/software/md5sum.exe reliably and
> apparently irretrievably results in 404 HTTP error.

Over the years, we've accumulated many files in
  /cache_for_lmi/downloads
If we've downloaded them once, in practice we never download them again,
so we don't detect the dead links. Thanks for finding this one.

>  Moreover, http://www.openoffice.org/dev_docs/using_md5sums.html#links
> listed as the place to find alternative download links, is itself
> unavailable. However looking at the old copies of this page, I see that
> it used to point to
> 
> http://www.pc-tools.net/files/win32/freeware/md5sums-1.2.zip#!md5!7e67b89a695bfbffc05b7ed2c38f927f
> 
> which is still available and so should presumably be used instead.

Apache links to that website, too:
  
https://svn.apache.org/repos/asf/incubator/ooo/ooo-site/trunk/content/dev_docs/using_md5sums.html#links
which is some sort of witness to its reliability, I suppose. But I
just feel a little uncomfortable with any "/files/win32/freeware/"
server. It seemed safe back in the days when we could scan '.com'
files in from magazines with a cauzin softstrip reader, but things
are different in this century.

>  However relying on the continued maintenance of these very old files
> doesn't seem wise in the long term and I think it would be better to either
> copy md5sum.exe to some place from which we could be sure it won't
> disappear or maybe compile our own version on the target system instead.

For the long term, we should either build it ourselves (perhaps as
part of lmi, since we already have part of it in git), or use a
different technology, because md5sum is so 1990s.

In the short term, it's not an obstacle for us, so I'll just flag
it for eventual action.

>  In any case, I think it would also be better to ignore the error from
> md5sum installation and install the other targets from this makefile, as
> md5sum is not absolutely required for building lmi itself (but only for
> "fardel" target, if I trust the comments in the makefile).

I think it's used also at run time (for end users who don't know
the Black Speech incantation), and we definitely have a unit test
or two that use it (though perhaps in a context where any program
at all would do, and we chose this one only because it seemed handy).