On Tue, Jul 28, 2009 at 04:35:31PM -0500, Shawn Walker wrote:
> Below is the second draft of a proposal for the adoption of a new  
> catalog format for the client and server.  As this is a revision of the  
> first format, and I believe I have addressed all noted concerns,  
> substantive feedback is requested by  COB of Friday, July 31st 2009.

> 2.1.2  Proposed Catalog Format

<...>

I like the changes you've made to the catalog format, but I still think
the updatelog format can be improved.

>     To enable incremental catalog updates, an "updatelog" will also be
>     provided as a single, merged file that can be used to incrementally
>     update any of the catalog parts.  It is composed of the following
>     files:
>
>     - updatelog.attrs.<logdate>
>         This file will contain a python dict structure serialized in
>         JSON (JavaScript Object Notation) format.  The metadata within
>         is used to describe the updatelog file and its contents using
>         the following attributes:

<...>

>     - updatelog.<logdate>
>
>         This file will contain a python dict structure serialized in
>         JSON (JavaScript Object Notation) format.

It seems unnecessary to have a separate updatelog.attrs file.  The
attributes should probably live with all of the others in catalog.attrs,
or in the updatelog.<logdate> file itself.  It better to have these
files be self-describing and I have some concern about the number of
round-trips the client will have to make to perform the download.

In the current model, the client makes one request and either gets a
304, a full catalog, or a stream of updates.  In the new model, the
client must download catalog.attrs before it can determine, what, if
anything, needs to be subsequently downloaded.  To get content, this is
a minimum of two round trips, with potentially pipelined requests.  If
the client also has to download the attributes for the updatelog prior
to performing a download of the updatelog data, we've introduced an
extra trip that's going to be a pain for users on links with high
latency.  It would be better if we could download a single attributes
file that describes all of the metadata we need to retrieve, and use
that to batch up a single subsequent multiple item request (piplined).

Thanks,


-j


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

Reply via email to