Hi, i think this is a good idea. The default output should show the last changed date. This could be very helpful for LIRs. Kind regards Eric
Erics wrote:
Hi,
i think this is a good idea. The default output should show the last changed date. This could be very helpful for LIRs.
It is easy for users who need this information to get it -- add the "-B" flag to your query. Also, most users of the database are not LIRs. Relying on the "changed:" attribute to know when objects were updated has the problem that there is no error checking on it. A user can put almost any "changed:" value that they want. I could update my person object today to have this value: changed: ripe-dbm@ripe.net 19901215 I think in the past people have suggested a "reliable" timestamp on objects in the database, but other people have argued against it. One suggestion was to add an optional attribute with a reliable timestamp that would get generated by the database. This way, if it is there you can trust it, but if it is not then you must rely on "changed:" attributes or do without any information. -- Shane
Hi, IMHO preferable solution is to modify "changed:" attribute format: changed: <e-mail address> with automatically generated timestamp. Such modification also allows exclude errors like "changed:" attribute order, updates with untouched "changed:" etc. Best regards, Rimas Janusauskas, Vilnius University Hostmaster On Wed, 11 May 2005, Shane Kerr wrote:
Date: Wed, 11 May 2005 10:13:31 +0200 From: Shane Kerr <shane@ripe.net> To: Erics <EricS@t-com.net> Cc: db-wg@ripe.net Subject: Re: [db-wg] Whois changes
Erics wrote:
Hi,
i think this is a good idea. The default output should show the last changed date. This could be very helpful for LIRs.
It is easy for users who need this information to get it -- add the "-B" flag to your query. Also, most users of the database are not LIRs.
Relying on the "changed:" attribute to know when objects were updated has the problem that there is no error checking on it. A user can put almost any "changed:" value that they want. I could update my person object today to have this value:
changed: ripe-dbm@ripe.net 19901215
I think in the past people have suggested a "reliable" timestamp on objects in the database, but other people have argued against it.
One suggestion was to add an optional attribute with a reliable timestamp that would get generated by the database. This way, if it is there you can trust it, but if it is not then you must rely on "changed:" attributes or do without any information.
-- Shane
Rimas Janusauskas wrote:
IMHO preferable solution is to modify "changed:" attribute format: changed: <e-mail address> with automatically generated timestamp.
The software, at least IRRd, does this. I never add timestamps. I also use nobody@aarnet.edu.au as the email address to avoid SPAM reports sent to the poor DB updater. Mark.
Mark Prior wrote:
Rimas Janusauskas wrote:
IMHO preferable solution is to modify "changed:" attribute format: changed: <e-mail address> with automatically generated timestamp.
The software, at least IRRd, does this. I never add timestamps. I also use nobody@aarnet.edu.au as the email address to avoid SPAM reports sent to the poor DB updater.
The RIPE Whois software also works like this. changed: shane@ripe.net Will become: changed: shane@ripe.net 20050511 If added today. -- Shane
On Wed, 11 May 2005, Shane Kerr wrote:
Date: Wed, 11 May 2005 13:50:36 +0200 From: Shane Kerr <shane@ripe.net> To: Mark Prior <mrp@mrp.net> Cc: Rimas Janusauskas <Rimas.Janusauskas@sc.vu.lt>, db-wg@ripe.net Subject: Re: [db-wg] Whois changes
Mark Prior wrote:
Rimas Janusauskas wrote:
IMHO preferable solution is to modify "changed:" attribute format: changed: <e-mail address> with automatically generated timestamp.
The software, at least IRRd, does this. I never add timestamps. I also use nobody@aarnet.edu.au as the email address to avoid SPAM reports sent to the poor DB updater.
The RIPE Whois software also works like this.
changed: shane@ripe.net
Will become:
changed: shane@ripe.net 20050511
If added today.
If... The key idea of my suggestion was do not allow "pour updater" ((C)Mark Prior :) set timestamps or ignore them (your previous message proves) Shane> A user can put almost any "changed:" value that they want. I could Shane> update my person object today to have this value: Shane> Shane> changed: ripe-dbm@ripe.net 19901215 Best regards, Rimas Janusauskas
On Wed, 11 May 2005, Rimas Janusauskas wrote:
Hi,
IMHO preferable solution is to modify "changed:" attribute format: changed: <e-mail address> with automatically generated timestamp.
Such modification also allows exclude errors like "changed:" attribute order, updates with untouched "changed:" etc.
Well.. I have had use of this feature, to be able to change objects without changing the changed-lines, when the ripe hostmasters wrongly changed a couple of thousand of our objects, several times a row. (during arin->ripe movements of old inetnums) So, I dont know, if ripe promise to behave, maybe this is a good thing.
Best regards,
Rimas Janusauskas, Vilnius University Hostmaster
On Wed, 11 May 2005, Shane Kerr wrote:
Date: Wed, 11 May 2005 10:13:31 +0200 From: Shane Kerr <shane@ripe.net> To: Erics <EricS@t-com.net> Cc: db-wg@ripe.net Subject: Re: [db-wg] Whois changes
Erics wrote:
Hi,
i think this is a good idea. The default output should show the last changed date. This could be very helpful for LIRs.
It is easy for users who need this information to get it -- add the "-B" flag to your query. Also, most users of the database are not LIRs.
Relying on the "changed:" attribute to know when objects were updated has the problem that there is no error checking on it. A user can put almost any "changed:" value that they want. I could update my person object today to have this value:
changed: ripe-dbm@ripe.net 19901215
I think in the past people have suggested a "reliable" timestamp on objects in the database, but other people have argued against it.
One suggestion was to add an optional attribute with a reliable timestamp that would get generated by the database. This way, if it is there you can trust it, but if it is not then you must rely on "changed:" attributes or do without any information.
-- Shane
-- Mvh /Fredrik ------------------------------------------------------- KTHNOC, KTH, S-100 44 Stockholm, Sweden +46 8 790 65 17 -------------------------------------------------------
On Wed, May 11, 2005 at 07:32:32AM +0200, Erics wrote:
Hi,
i think this is a good idea. The default output should show the last changed date. This could be very helpful for LIRs.
The fact that changed attributes are now suppressed by default is they contain email addresses. I think at some point of the discussion we mentioned the possibillity to change the attribute syntax to a NIC-handle, but it would require an enourmous amount of updates to fix all exisiting objects. Have no idea on how easy it would be and although I don't like the idea of changing the whole lot again in such a short time. But it might be an alternative. Grtx, MarcoH
participants (6)
-
Erics -
Fredrik Widell -
MarcoH -
Mark Prior -
Rimas Janusauskas -
Shane Kerr