Danek Duvall wrote:
> On Tue, Jun 24, 2008 at 10:59:38AM -0700, Joe Di Pol wrote:
> 
>> For example would it be like:
>>
>> add file ... path="foobar/doc/CHANGELIST.txt"
>> add set name=my.changelist value="foobar/doc/CHANGELIST.txt"
> 
> You could do that, but that potentially would require a second lookup in
> order to retrieve the file.  Assume that in the future you'd be able to get
> back the "my.changelist" set action of a package (or multiple packages)
> without having to get the entire manifest, so then you'd have to do a
> search to find out the hash, and then you could retrieve the file.

Gotcha. Any idea how far in the future this would be?

> 
>> Or (gulp) this:
>>
>> add file ... path="foobar/doc/CHANGELIST.txt"
>> add set name=my.changelist set value="file/0/<hashval of CHANGELIST.txt>"
>>
>> The later doesn't seem very useful since you don't know the URL at
>> package publish time.
> 
> Why don't you? 

OK, so you could. I guess in this case it would look like:

add set name=my.changelist set value="<hashval of CHANGELIST.txt>"

And the client would expand it to a URL at runtime. Looks like
something I can experiment with now.

But is the hashing algorithm a stable interface? If I'm some random
package maintainer, can I always depend on using the SHA-1 hash?

The first approach seems more robust and friendly for a package maintainer
(albeit a bit more complicated and costly for the client).

Joe
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to