Re: [Gnu-arch-users] Tlaa on wndows - new pathcompression scheme

[John F Meinel Jr]

Just as a side note, I agree. I don't think we want to switch to the 
VFAT names. You still have the problem that you can't delete a directory 
if the names get too long. I suppose you could look at directories then 
and have some indication of what it is, but I also know that my project 
tree gets a little deep (more like 80chars or so.) So that does cut into 
the max length available. I think Johannes' work around for exporting a 
network drive and re-mounting it is interesting, but I don't think it 
should be required to be able to use TLA.
John
=:->

Parker, Ron wrote:

> > From: address@hidden [mailto:address@hidden
> >    
>
>  
>
> > Someone on this mailing list wrote that pathcompression
> > appears to work, but that the mapping to numbers is kinda
> > ugly... or he didn't realy like it but could live with it.
> >
> > First reaction was to think "shoo, go away fool"
> >    
>
> It was me.  I posted from a different address, but I did say that.  I also
> said it was faster to do a number-to-string lookup than it was to do some
> sort of string-to-string mapping.  
>
> If the VFAT short name hack is used directly via the Win32 API, it solves
> one class of problem, i.e. get tla working on Win32.  It does not
> necessarily solve the Microsoft-application-X-crashes-on-deep-paths problem
> which afflicts Explorer and many others, as they always opt to use the long
> path form, unless explicitly given the short form.  
>
> However, if you are talking about some generic solution similar to VFAT
> short names that might be workable across platforms.  It would still be
> slower than a number-to-string mapping, but it would be somewhat more
> understandable to the user as well.  Overall I think it would be a win.
>