Hi John, On 2011-11-28 21:50 John C Klensin said the following: > > > --On Monday, November 28, 2011 21:42 +0100 Henrik Levkowetz > <henrik@xxxxxxxxxxxxx> wrote: > >>> One small suggestion, partially prompted by my attempts to >>> convert PDF and Postscript RFCs to PDF/A: when the converter >>> cannot or does not succeed in producing valid PDF/A, could >>> that fact be logged in some accessible place? >> >> I would if I could; the problem is that the converter produces >> a pdf document without giving any indication that the result >> won't validate as PDF/A > > That is what I believe is called "broken". I hope someone can > make a problem report. Yes. I suspect the authors of the software aren't aware of the problem. Robinson Tryon has now filed a ticket with LibreOffice about it, though; and I've provided him with another sample which fails validation. >> -- it's only when trying it with one >> of the online validators that I get a message that it fails >> validation. If at some point I find a (free) command-line >> validator, I can however easily run it on the existing corpus >> and get the desired information. > > If the best (or only) way to handle this is via pos-production > validation, then we can presumably do that at any time... and > validators will probably get better over time too. Ack. > Suggestion withdrawn. > > thanks for considering it. Certainly! >>> Also, mostly out of curiosity, does "unoconv" produce profile >>> 1a or profile 1b? >> >> The validator tells me that the files which validate does so >> against both A-1a and A-1b, which if I understand things >> correctly indicate that it's 1a-compliant, since 1b is a >> subset of 1a. This also matches what libreoffice is supposed >> to produce. > > Yes. Without going down that rathole, I think (haven't studied > the issue closely enough to be confident) A-1b would be good > enough for our purposes so it may be vaguely possible that the > files actually are 1b-conformant. Not worth worrying about > until and unless we are ready to get really serious about this, > establish specific norms, etc. I don't see that happening > really soon despite the recent discussions. You may be right. Best regards, Henrik _______________________________________________ Ietf mailing list Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf