Is this really valid? Your spice client is not the only one using these headers, please see the Looking Glass project.

They are using the wrong enumeration, correct ones are SPICE_LINK_ERR_xxx. Can you open a bug to Looking Glass ?

Even if the enumeration is obsolete, we should not pull the plug like that on external users, I'd revert and release 0.12.15.1 (or whatever version works)

I don't agree, the bug is in the external code. Opened a pull request to fix the issue https://github.com/gnif/LookingGlass/pull/126.

I am the author of Looking Glass, thanks for the patch. I was unaware of the rename as the Spice client in LG was written less then a year ago based on the Spice PDF documentation which doesn't use this new enumeration. Perhaps a bug should be opened against the PDF document for being out of date.

Even if the external code has a bug and should not be using these enums, we should not break it, and then blame our users. So yes, the external code needs to be fixed, but this also shows that this removal was premature.

On the other hand we would never knew that the documentation was wrong. Waiting other 5/10 years won't have helped.

@geoff , where did you find the PDF ? The PDF I can find uses RED_ERROR_xxx constants which are still wrong but better than the client ones.

@fziglio it was the same PDF that you are seeing. https://www.spice-space.org/static/docs/spice_protocol.pdf

I wrote this code quite some time ago and my memory is a little fuzzy, however I did make the natural leap of grepping for the error defines based on the PDF which turned up the deprecated header.

Well, they seems similar to the RED_ERROR_xxx ones but the values are also different so using them would result in bad feedback to the user. It would be great to update the PDF, even better to convert to some markup/down (I think there are old patches about :-( ) and update. Thanks