On Sat, Nov 08, 2008 at 05:34:43PM +1100, Sisyphus wrote:
> No - it won't be those actual names.
>
> The 'abcdef.....' section will be replaced with the md5 digest (or relevant
> part thereof) that Inline has calculated for that specific Inline file.
[snip]
Great, that makes more sense. Thanks!
> How small a limit are we looking at ? Are there situations where files like
> 'script_pl_2f03a1cb102f1314deafda5f929b44e3.inl' would be disallowed ? If
> so, then I guess now (rather than later) is the time to address that issue.
I was actually thinking of Windows, with its semi-hidden filename
truncation to 8.3 characters - especially likely to cause problems if
the names had a common prefix. But if the existing behaviour is to name
the file script_pl_{md5 hash}, then we're already running into this
issue and surviving it, so it doesn't matter.
Of course, more primitive systems exist, with actual hard character
limits - very early Unix, DOS, Multics etc - but I'm guessing Inline
isn't supported on any of them?
Miles
--
A less powerful language forces me to do things the way the language
wants me to do it. For every transformation of my solution that I need
to do before getting it in code the more likely it is that I'll make a
serious mistake. A powerful language meets me rather than forces me to
meet it.
-- KayEss, on Reddit
