:-O Looks ok on a quick test. CoolSeeing as it's you...
(Update pushed - though done in a little bit of a hurry.)
But I've a bug to report. After a little discussion on Discord, this does seem to have been affecting multiple people over multiple revisions of the bridge...
Copying a bunch of files, or a folder full of stuff, using the RISC OS filer... from, say, RAM: to NET: ... it is observed that the access permissions get 'messed up'. But I think I can show an example where this behaviour is at least consistent, repeatable, and may have some inkling as to a factor causing the effect: We take as an example (where I first hit this as a real problem) the excellent Econet Elite archive from MarkMoxon. The screenshot below shows the contents of this as unzipped to a local RAM drive. Note the WR/wr. The right side shows a crispy fresh 2.1-pre Pi4 server where the files have been copied to.
Note that it seems that the files that are ok are all "small" in size - under 4KB or so. I'm wondering if this is about the block size that filer is using for its OS_GBPBs or something - whether files copied all at once are perhaps ok, but those that are created, then updated, get goosed? I'm not sure... but it definitely looks like a pattern to me, and is repeatable.
Note also some files have only lost their 'w' not the 'r' - these appear to be perhaps 'medium size' files ?
This issue caused a bit of headache wondering why elite would launch, then die, but it's because some files were not accessible by the elite user (no global read permission on the files) when accessed from the pi via Econet.
Statistics: Posted by IanJeffray — Wed May 15, 2024 1:07 am