hi guys,
We are getting ready to put the 6.0/updates into place. The prep work brought up a question that needs addressing.
There are quite a few different tools that seem to scrape the centos-announce output in order to get update metadata; therefore should we announce every update since the 6.0 release date upstream ? we might be looking at around 5 million emails being generated outbound...
if not, what are the other options / ideas ?
- KB
On 07/04/2011 08:08 PM, Karanbir Singh wrote:
hi guys,
We are getting ready to put the 6.0/updates into place. The prep work brought up a question that needs addressing.
There are quite a few different tools that seem to scrape the centos-announce output in order to get update metadata; therefore should we announce every update since the 6.0 release date upstream ? we might be looking at around 5 million emails being generated outbound...
if not, what are the other options / ideas ?
- KB
If people know it's coming, and if it comes in a flood, then concerned people can write a filter to redirect the flood to their trash.
Personally, I am a fan of too much info when the alternative is risking important data being missed. Just pre-define a date and time that the flood will start. Perhaps tell people to watch for a particular subject string that will indicate when the flood is complete.
Perhaps: ==== [ CentOS-devel ] == Flood Start == <announcements> [ CentOS-devel ] == Flood End == ====
A suggestion. :)
On 7/4/11 8:10 PM, Digimer wrote:
On 07/04/2011 08:08 PM, Karanbir Singh wrote:
hi guys,
We are getting ready to put the 6.0/updates into place. The prep work brought up a question that needs addressing.
There are quite a few different tools that seem to scrape the centos-announce output in order to get update metadata; therefore should we announce every update since the 6.0 release date upstream ? we might be looking at around 5 million emails being generated outbound...
if not, what are the other options / ideas ?
- KB
If people know it's coming, and if it comes in a flood, then concerned people can write a filter to redirect the flood to their trash.
Doesn't everyone use the digest version on the announce list anyway? Or at least everyone who would be bothered by a flood of messages?
On 07/05/2011 05:01 AM, Les Mikesell wrote:
On 7/4/11 8:10 PM, Digimer wrote:
On 07/04/2011 08:08 PM, Karanbir Singh wrote:
hi guys,
We are getting ready to put the 6.0/updates into place. The prep work brought up a question that needs addressing.
There are quite a few different tools that seem to scrape the centos-announce output in order to get update metadata; therefore should we announce every update since the 6.0 release date upstream ? we might be looking at around 5 million emails being generated outbound...
if not, what are the other options / ideas ?
- KB
If people know it's coming, and if it comes in a flood, then concerned people can write a filter to redirect the flood to their trash.
Doesn't everyone use the digest version on the announce list anyway?
No
On Tue, 05 Jul 2011 01:08:12 +0100 Karanbir Singh mail-lists@karan.org wrote:
hi guys,
We are getting ready to put the 6.0/updates into place. The prep work brought up a question that needs addressing.
There are quite a few different tools that seem to scrape the centos-announce output in order to get update metadata; therefore should we announce every update since the 6.0 release date upstream ? we might be looking at around 5 million emails being generated outbound...
if not, what are the other options / ideas ?
Not that I know the particulars on how these things are generated, but perhaps output to a text file and then 1 email that would point to a link containing that file? Or an xml file, or whatever format would suit the occasion. Then even if each one has to be done individually, the output could be appended >> to a specific location, hopefully?
Or, if even that would generate too many lines for one linked file, could it be parsed alphabetically or something and split into a few smaller files, with appropriate links on centos.org or the wiki or something.
On 5.7.2011 02:08, Karanbir Singh wrote:
There are quite a few different tools that seem to scrape the centos-announce output in order to get update metadata; therefore should we announce every update since the 6.0 release date upstream ? we might be looking at around 5 million emails being generated outbound...
if not, what are the other options / ideas ?
Not sending them could break those tools. People who dont want to have the mails are free not to subscribe. Subscribed people who really care could temporary turn off the delivery of mail which is possible with mailman.
On Tue, Jul 05, 2011 at 01:08:12AM +0100, Karanbir Singh wrote:
There are quite a few different tools that seem to scrape the centos-announce output in order to get update metadata; therefore should we announce every update since the 6.0 release date upstream ? we might be looking at around 5 million emails being generated outbound...
Please *do* send the update notices. It'll help the automated processes based around it, and also will demonstrate that hey, this is *not* just swept under the rug.
Please send updates. That being said, I remember months ago you had talked about perhaps offering a better method of data dump from CentOS than the email method for automation purposes, what ever happened to that?
-Blake
On 7/5/2011 10:15 AM, Blake Dunlap wrote:
Please send updates. That being said, I remember months ago you had talked about perhaps offering a better method of data dump from CentOS than the email method for automation purposes, what ever happened to that?
Seems like something that would be perfect in an rss feed, although programs that processed it would have to track their own timestamps and be careful not to do something stupid if they lose their place from being reinstalled or having a backup restored. For the 'human readable' side, pointing google reader or a similar feed aggregator at it should be fine, and turns it into something where new items are easily visible when you look without creating clutter or requiring visits to lots of individual sites to check for updates.
On 07/05/2011 04:49 PM, Les Mikesell wrote:
On 7/5/2011 10:15 AM, Blake Dunlap wrote:
Please send updates. That being said, I remember months ago you had talked about perhaps offering a better method of data dump from CentOS than the email method for automation purposes, what ever happened to that?
Seems like something that would be perfect in an rss feed, although
repo-rss does that now, we can publish that now with very little effort.
- KB
On 07/05/2011 04:15 PM, Blake Dunlap wrote:
Please send updates. That being said, I remember months ago you had talked about perhaps offering a better method of data dump from CentOS than the email method for automation purposes, what ever happened to that?
that sort of changed in scope - its still in the pipeline and I hope to have some usable code soon. Atleast enough of the framework that other people can then take it up and add functionality etc.
the plan that I've been working towards is to have a REST api that can be called from whatever service / consumer / frontend and have information published there. the metadata that is needed for something of this nature is already being aggregated ( stuff like changelogs and cve's etc )
- KB
-
Blake Dunlap -
Cia Watson -
Digimer -
Karanbir Singh -
Les Mikesell -
Manuel Wolfshant -
Markus Falb -
Matthew Miller