wrote:
On Wed, 10 Mar 2021 13:08:02 +0000
Recliner wrote:
On Wed, 10 Mar 2021 11:53:02 +0000, Roland Perry wrote:
Anyway, he's on the case, and has submitted a bug report to Apple, and is
also trying to creating a workaround to the iOS bug, on the assumption that
a fix won't come quickly:

"Therefore I guess this issue must be relatively new, but you might be
right and this was already a bug in iOS 13 (which was very buggy in the
first place) and is not yet fixed. Nevertheless, I’ve sent a bug report to
Apple, so hopefully Apple will fix this some day. And because Apple is lazy
in fixing bugs which are not getting public interest (list security bugs),
I’m also writing a replacement for the buggy iOS call, so I can work around
this iOS bug. Unfortunately Apple has not documented the file format
(IMHO), so this might need a short while to test everything..."

I wonder if (a bit like the infamous Covid test results being stored in
an Excel spreadsheet) a quick fix would be to write that data in two
files, and glue it back together after they've been read back in.

Yes, that's one possible workaround, but if going down that route, you would
write n small linked files, not two
medium-sized files. It's how large binary attachments are handled in
newsgroups. Anyway, he seems to have a tidier
work-around proposal — let's see if it works.

I'm struggling to imagine how something as fundamental as writing files could
have a bug in it which hasn't been noticed before.



This is how he described the problem:

The problem seems to be that the system call that is used to load the
thread structure of a group from the file system is unable to do this, even
though the iOS system call to save this structure to the file system has
done so without any issue. The problem seems to happen only for extremely
deep nested threads. When saving these deeply nested structures iOS reports
no error, but when reading them, iOS reports an error.