On 25/03/2015 21:23, Karl Berry wrote:
> saying words to the effect 'You better have a matching set of
> files'.
>
> I agree.  And my point is that a user cannot know if the files are
> in fact "matching" unless you specify actual information about the
> files used to build.  A random version string like "7.0.0" is not
> enough to technically verify.  A checksum (ideal), a file size (good
> enough in practice), something that is specific to the actual content
> is needed. Otherwise it's just a guess.
>
> Regarding the name: I like unicode-letters.def.
>
> thanks, karl
>

I don't object to adding a checksum (especially as Joseph's doing the
work;-) but I expect for most people it's just as easy to check that
from the version number. If whatever-its-called.def says it is generated
from version 7.0.0 people can just do as you said you did in your
original message, go to version 7.0.0 of the UCD, get the files and
generate it themselves. If they get a different answer, something is wrong.

If there is a checksum in the file the procedure is more or less the
same, get the right version of the UCD data files (for which the version
number is needed anyway) generate a new version of the .def
file, the only difference is that then you can compare the generated
checksum comments rather than use diff or fc etc to compare the files.


David



_____________________________________________________________________________________________________________________________________________________

The Numerical Algorithms Group Ltd is a company registered in England and Wales with company number 1249803. The registered office is:

Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.



This e-mail has been scanned for all viruses by Microsoft Office 365.

______________________________________________________________________________________________________________________________________________________