Latest word on DR-16-0014

Francis Cave francis at
Mon Jun 26 23:32:03 CEST 2017

I will have to defer to greater minds than mine on this question. However, it seems to me that Courtenay is wrong: an rsid doesn’t have to be completely randomly-assigned, so long as the seed is unique. A combination of MAC address and current universal time would, for practical purposes, provide a reasonable guarantee of uniqueness. 






From: Rex Jaeschke [mailto:rex at] 
Sent: 26 June 2017 21:57
To: SC 34 WG4 <e-SC34-WG4 at>
Subject: FW: Latest word on DR-16-0014


Feedback from the submitter. Rex


From: Courtenay Inchbald [mailto:c at] 
Sent: Monday, June 26, 2017 7:49 AM
To: Rex Jaeschke <rex at <mailto:rex at> >
Subject: Re: Latest word on DR-16-0014


Dear Rex,


Thanks for the update. I presume that the text you are proposing is the final version in the document, i.e. the edits to my new short version, not the edits to old longer version. If so, I have no objection to the rules. I am sorry my enquiry created so much work.


However, I must point out that if you do not specify random generation of Revision Save IDs in the second rule, some readers will presume that there is some way, or that they should devise some way, of checking all branched copies of a document, or a central database of their Revision Save IDs, before generating a new Revision Save ID. Since this is not practical, the only way to minimise the probability of a duplicated Revision Save ID is random generation from the values not already listed in the rsids element. Any other method will not minimise the probability of duplication. I make this point because, in my experience, even some of the best-educated people are easily confused by probability.


Courtenay Inchbald  | T +44 20 8673 5640 | M +44 771 858 3535 | 54 Elmfield Road London SW17 8AL | c at <mailto:c at> 


On 26 June 2017 at 08:01, Rex Jaeschke <rex at <mailto:rex at> > wrote:



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the sc34wg4 mailing list