[m-dev.] for review: warn about problems with `make install'

[Warwick Harvey]

Fergus wrote:
> .INSTALL.in:
> 	Warn about the problems that can occur if `make install'
> 	gets half-way.

[...]

> +#	   BEWARE: if something goes wrong in this step, e.g. running
> +#	   out of disk space, and the installation has already gotten
> +#	   past installing the documentation, and started installing
> +#	   the different grades of the libraries (in particular if
> +#	   `grep install_text make_install.log' prints out something
> +#	   like `# install_text install_ps') then in general it is NOT
> +#	   sufficient to simply rerun `make install'.  Instead you
> +#	   may need to start from scratch again.

Is it worth trying to provide something better than this?  If I understand 
the installation process correctly, the existence of tmp_dir subdirectories 
indicates a failed partial install, and moving their contents back to the 
parent directories is all that is required to restore things to a reasonable 
state.  (It shouldn't be necessary to delete any files in the parent 
directory [other than those clobbered], right?)

If this is true, then a script could be written to check for such 
directories and move the relevant files.  If such a script were written, I 
presume we would be able to persuade mmake to run it at the start of the 
installation process?  Would that work?  Would it "solve" the problem?

If so, I'm happy to have a bash at the script.

>  # Step 4.  Check the permissions on the installed files.

I always thought this bit sucked, too.  (I get too many complaints from Greg 
Denehy that there are files he can't read after I've installed a new version 
of the compiler ;-).  I'll see if I can find some time to look into this one.

Warwick