>> Compress a 60k JAR of HTML to 6k and insert it.  Then, retrieving any of
>> the HTML pages gets you all of them.
>> Insert the small images that appear on more than one page as a group
>> again, which may be 150k or so, and insert it so that retrieving any of
>> those common images gets you all of them for the site.

>All this is good. We just have to implement <ZIP key>//<file in ZIP>,
>and all of this becomes possible.

well, we had the top of this mail now 100 times.. maybe everybody now realizes that 
bundling files which belong together might lead to a good idea....

---~~---

but matthew, why your approach with ZIP@ ?
why not use oskar's approach with metadata information which makes *far* more sense to 
me.
ZIP@ is stupid, because the data contained within the archive does not need a special 
key - it's a file just like any other file, so no special key is needed

> OK. Proposal:
> Currently manifest sites use Redirect.Target=<some key> or DateRedirect
> (which has .Increment, .Target and .Offset).
> Add ZIPRedirect:
> ZIPRedirect.Target=<some key, which is a ZIP>
> ZIPRedirect.Filename=<filename within zip to extract>

redirections may work but i think it's the wrong approach, too. it's simply not 
transparent enough. so every file which resides within the archive must be 
indexed with a ZIPRedirect.Filename param, which will simply be a too big workaround 
for embedding archives

my idea would be based on metadata information, too, but would look like this:
>>
Version
Revision=1
EndPart
[...]
Document
Redirect.Target=freenet:SSK@some/some//mycommon.zip/next.gif
Name=next.gif
Info.Format=image/gif
EndPart
[...]
Document
Redirect.Target=freenet:CHK@blablaarchivefiledata
Name=mycommon.zip
Info.Format=application/x-freenet-site-archive (or application/x-zip or whatever 
archive format fproxy/etc should be able to understand)
Info.UseAsABundle=true
EndPart
[...]
End
<<

notice here the additional field "Info.UseAsABundle", set to "true" (default "false").

this enables the client to see that the file the "Name" fields describes is an archive 
(and not a "directory") which can be opened to retrieve files like 
"mycommon.zip/next.gif" from within the archive.

everything else on the metadata remains untouched, no new key types are needed.

[[[ i'm not very good at metadata discussions, but maybe this is a way you can gather 
some clues from? ]]]

---~~---

>It could be worth discussing whether fproxy should be caching the data 
>contents itself, or reextract the bundle every time (counting on the 
>DataStore to cache it).

we have a temp directory, use it for extracting the archive (maybe if >n files are 
going to be retrieved from the archive, otherwise do not expand the archive 
into temp)
this will leave some files lying around, i know.

so maybe it's wise to extract the files into the datastore (using their own CHK, not 
the archive's) so the first look can go into the DS, if the requested file is 
not present, look into the archive, and possibly extract the file found into the DS, 
where it is cached when the file is requested the next time.

---~~---

comments?

____________________________________________________________

The information transmitted is intended only for the person or entity to which it is 
addressed and may contain confidential and/or privileged material. Any 
review, retransmission, dissemination or other use of, or taking of any action in 
reliance upon, this information by persons or entities other than the intended 
recipient is prohibited. If you received this in error, please contact the sender and 
delete the material from any computer.






_______________________________________________
support mailing list
[EMAIL PROTECTED]
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/support

Reply via email to