Less stringent column name checking in __strip_relcond

[mzealey]

If (like us) you are using some strange syntax for column names the \w+ regex
breaks returning () which leads to an incorrect hash.

[ribasushi]

@mzealey I am preparing an alternative changeset which addresses the above in a much more substantial way. It is going to be ready by tomorrow, at which point I would like to have you test it.

In the meantime: given your report made me revisit the entire codepath I would like to add you to the AUTHORS list, could you please let me know what line would you prefer given the format?

Thanks!

[mzealey]

Thanks. Just Mark Zealey <mark@dns-consultants.com> is fine

[ribasushi]

Just Mark Zealey <mark@dns-consultants.com> is fine

Would it be ok to add your IRC nick as well? Or was that someone else I spoke to on IRC about this?
jalh: Mark Zealey <mark@dns-consultants.com>

[mzealey]

ok

[ribasushi]

This has now been addressed by 86be9bcb9 (part of a much larger rework, prompted by a different set of problems). Please check out latest master (89d5e36fd) and see if the problem still persists (it is possible that the now-official resolve_relationship_condition() has flaws related to parsing your use-case)

I will keep the PR open for some time as a reminder it still needs input from you.