About a year ago there was a issue that hit the news about a potential security leak in Intel processors.  The problem had to do with one process reading the cache of another process.

The correction, which for most was done via software, was notable in that it slowed down the systems quite significantly.  Perhaps this is what you are seeing.

Here is a link:  https://www.extremetech.com/computing/261420-early-data-shows-linux-update-fix-intel-security-flaw-hits-performance-hard

You might want to check the size of the disc cache and buffers on the various platforms.

Try:   free -m  (The -m indicates output MB)

The newer operating system may be using more memory (not uncommon) resulting in less space in memory to buffer data.

TCB(88) is the version of the WindX program not the version of PxPlus.  The WindX version doesn't change unless there is new functionality added to the WindX<-->Host protocol.

Our utilities use the value in TCB(88) to determine what options on the protocol are available.

Make sure you are running with Turbo mode enabled (system parameter 'TU' enabled).  Generally it should be on when using Nomads.

Try removing the WindX INI file which contains the authorization information.

It should be:

C:\Users\<username>\AppData\Roaming\PxPlus\windxcfg.ini

By default the PasteFilter is off on ALL input that provide for multiple lines of input and takes the setting from the 'PasteFilter" option for single line inputs (default is on)

When I created a panel here with a multiple line high input and loaded the clipboard using:

CLIP_BOARD WRITE "Line 1"+$0d0a$+"Line 2"+$0909$+"tab"+$0d0a$+"Line3"+$0d0a$

If I then pasted this into the input I got three lines.

However we have noted that for some reason there seems to be a difference in "How" you paste the data.  If you right click and paste the filter seems to always apply, however using the keyboard paste CTRL-V the filter does not unless enabled.

We will see if we can determine the reason for the discrepancy.

We looked at the logic in memory files and could not an error 2.  The only thing we can think of, assuming its really a memory file, might be if you had a pending IO to a file that returned an error 2 due to space exhaustion or the like.  Some file IO operations are buffered to improve performance.  Its possible when flushing a ending IO operation in background we received an error condition.

It could also be some form of embedded IO for the memory file that itself had the error 2. 

Can you also confirm you are running PVX 15.10 -- To the best of my knowledge there never was a PVX 15.10 -- there was a PxPlus 15.10 but not PVX 15.10.

Stéphane,

For us we host the SVN server in-house -- actually quite literally in-house as the server resides off site at our home thus serving not only as the repository but off site backup for all the source.

We back the repository up nightly to both a server in our office and to one of our web site servers which is at a hosting provider (and get daily emails confirming backup was successful -- so at least someone sends me mail on a daily basis )

This setup has worked well for the past 15 years (that is when PVX Plus was founded) and, apart from the odd time when the internet was down, it has provided us easy and secure access to the SVN repository for all our staff. 

Setting up the SVN server on a Linux system was pretty easy to do and took about a hour what with installing and configure the software.  We firewalled the router and network plus provided access keys (SSH encryption certificates) for each of the developers.

We opted to have a private server as it is less a target than a centralized hosting service provider or things like GITHUB might be.  If someone wants to hack into a server they are more likely to target larger sites as the reward would be greater.

As I said, this has worked well for 15 years so my personal recommendation would be to do it in-house as opposed to trying to deal with 3rd parties for setup, backup and ongoing control of user access.