Concerning: "... so what is it that ties the EPO to Amyuni ?"
Well, the EPO have opted for it as it allowed to set-up an installation version that would preset the Annex F requirements and disable any deviations from that. At least that was and still is the idea.
Of course, any other PDF converter is likewise suitable if configured according to the Annex F requirements which are well documented e.g. in the FAQs: http://docs.epoline.org/onlinefilingdoc ... ppo_EN.pdf
But it is true that with v4 last September, the PDF verification mechanism has been made stricter to adhere to Annex F which in some aspects turns out to be too strict, e.g. Annex F limts the page size to A4.
Well it does not say anything about tolerances and thus even a 0.001 millimetre now lead to a rejection by the OLF system - this is currently being reviewed and will be changed again such that scans from machines that were not able to produce strict A4 will be accepted again (too an extend, of course, 5% is the current idea). This is where currently users perform a workaround by re-generating the scanned image with a PDF converter that produces strict A4 measures.
And if you think that all PDFs follow the same standard, well think again:
1. not all PDFs are 100% standard conform,
2. the standard allows for some variation which any verification mechanism must then be able to cope with - this is were the problem lies.
But in hindsight, this is a quality issue which needs addressing, of course, and any user feedback is helpful in this respect.
Allow me to digress a bit:
The long term goal will be to get away from this paper imitating format PDF to a more structured and data orientated text format but although the XML standard for application-body exists already, there have been several attempts in offering tools to create the patent specification in that format (PatXML, PCT-SAFE Editor, HTML2XML converter by JPO, etc.), none has yet had the desired impact that it can fully replace the word processors with all their feature richness. If anyone comes up with a simple way of converting ODF and MS Open XML into the application-body.dtd compliant form prior to attaching it to online filing while ensuring that the user has enough confidence that the converted format contains the relevant text and data - this would be killer application.
We will improve - eventually