[Virtualacorn-list] RO 5.20 on VRPC

T.O.M.S. admin at toms12.plus.com
Thu Jan 9 19:59:44 GMT 2014


On Wed Jan 8 12:14:02 GMT 2014 David J. Ruck druck at druck.org.uk :

>> A wish for >256MB RAM is an interesting point though. In practice
>> we never get even close to the limits with 256MB as we find that
>> handling such enormous files would become so unwieldy that it's
>> much better to split the files into smaller and more manageable
>> sub-sets. That also avoids the risk of hitting the buffers of the
>> 26-bit limit of (typically) 30MB.

> The subject line indicates running RISC OS 5.20, which doesn't
> have the 26bit limit of 28MB for the application slot, and allows
> a single task to use all the free memory in the machine in its slot.

Yes indeed. I mentioned the 26-bit limit of RISC OS 4 intentionally,
as a reminder of what had already been discussed on this thread back
in August 2013 (also on c.s.a. and the VirtualAcorn Forum in the last
couple of years).

Until this present discussion got nicely going, all previous threads
concluded that there was seemingly no significant advantage *in
practice* to being able to run RISC OS 5.xx under VirtualRPC.

It's interesting that you mentioned stitching images together as the
best example I could give of hitting the 26-bit 'buffer' under RISC
OS 4 is when we stitched 52 images together (necessarily under
Windows) and which resulted in a 44Mb TIF file. Even if VirtualRPC was
running RISC OS 5, on a far more powerful host PC than Iyonix,
massaging and annotating the enormous bitmap in (say) Artworks would
have been quite impractical.

Hence my comment about the value of splitting files into more
manageable portions. Sub-dividing the 44MB TIF into 4 equal segments,
using David Pilling's DPScan for Windows, then saving them out as A3
drawfiles to import direct into Draw, was a trivial task. (Thanks
David!)  Butting the 4 finished segments back together as a monster A1
chart and converting it to PDF, all ready for printing, was equally
trivial using Xara under Windows.

In other words, IMHO the 26-bit buffer of RISC OS 4 is not (for us) a
limitation *in practice*, but again I find it interesting that both
you and Roger D have now identified circumstances where it could be
restrictive (running ported command line processes; also running
DataSynch and even Netsurf). This is a whole new slant on the
much-discussed topic. Thanks to both of you for that.

Alex Hamilton
pp T.O.M.S.






More information about the Virtualacorn-list mailing list