On Tue, Oct 5, 2010 at 10:49 PM, Hisham <[email protected]> wrote:
> First, there's the issue to how to interpret arguments (the third
> argument is, by default, the version number). Myself, I would just use
> 'for' or 'xargs' in the shell, but I can see the convenience of having
> something like this...

That did occur to me, and there's no easy way around it, maybe an
extra flag.  While we're adding imaginary flags, there could be one
which checks the current package version against the latest package
version and doesn't try to re-install.  That is appropriate behaviour
for 'luarocks make' of course, but it's generally a waste to
re-install something that's identical.

If I do a prototype, it will be a separate command at first.

> If the error is reproducible, one can just do "luarocks xxx &>
> error.log" anyway.

This is true, except for some shell-challenged people ;)  But what's
irritating there is that LR is being too good at rolling back,
including deleting the downloaded stuff.  Being at the periphery of
the net, at the end of a few thin undersea cables, I'm very aware that
there's a fat latency involved in getting something downloaded.   So
it would be cool if a failure to install would not delete the
downloaded stuff - yes, I know, cache management is hard - this is
just a bit of practical dreaming.

>> which pull in a whole bunch of common packages as dependencies.
> That's what I would do if I wanted to implement that (or write a
> script that did the sequence of commands)...

This makes sense - make up a fake rockspec and populate it with the
required packages as dependencies.  Then, if we are being thorough,
remove that fake package afterwards.

steve d.

_______________________________________________
Luarocks-developers mailing list
[email protected]
http://lists.luaforge.net/cgi-bin/mailman/listinfo/luarocks-developers

Reply via email to