[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: TMS comments

Hi RaphaŽl.

On Wed, 4 Sep 2002, RaphaŽl Quinet wrote:
> What I am proposing is a bit different: the applications supporting
> the TMS would never write thumbnails in these shared directories.  
> The new thumbnails would always be created in the users' home
> directories, where write access is (usually) not an issue.  But when
> reading thumbnails, the applications would always check the local
> (shared) directory first, then the user's home.  If there is no local
> thumbnail for some image or if it is not valid, then we are back to
> the normal case proposed in the TMS 0.6.
The TMS uses MD5 hashes of absolute URIs for identifiying an appropriate
thumbnail. If you put a .thumbnails dir on a CD, your users benefit from
it only if they use the same mount path than the creator of the CD
did. The same applies to NFS mounted directories.

You see, it requires more basic changes in the TMS. It could be done the
following way: You have to decide if there is a local .thumbnails dir, if
so then generate a relative URI hash and check it against the local
.thumbnails dir. If this fails, then calculate the absolute URI hash and
check against the global .thumbnails dir in your home.

IMHO such special cases should be avoided in the TMS. Although I see the
practical benefit of your proposal. But we won't get it without the
sketched changes AFAICS.



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]