Works with "svn, version 1.9.7 (r1800392)" on OpenBSD.

Masscheck confirmed working fine on my CentOS 7 VM running SVN version 1.7.14 for the past 10 months.

For those of us who want to use the method here: https://wiki.apache.org/spamassassin/CorpusNightlyScript this is a bug. I know it's regarded as an alternative method, but it's still valid as far as I am aware. This extremely simple patch doesn't break anything for anyone using the default method but it does enable those of us who wish to use that method to perform masschecks without manual hacking.

See new features section: https://subversion.apache.org/docs/release-notes/1.7.html
Also noted: https://en.wikipedia.org/wiki/Apache_Subversion#Limitations_and_problems

The bug is valid, if you don't want to fix it then WONTFIX is more appropriate.

Sorry about that.  The problem was not clear until that link was provided. I didn't know there was an alternate way to run the masscheck besides the https://wiki.apache.org/spamassassin/NightlyMassCheck.  I see the CorpusNightlyScript now at the bottom of that page.

What is the difference with the CorpusNightlyScript way?  I will be glad to fix this once I understand what is needed.  I am familiar with the automasscheck-minimal.sh but not with this alternative way.

I believe the biggest difference, and the reason I personally prefer using it, is that it is sh rather than bash - as a FreeBSD user bash is an additional dependency that requires installing and I prefer not to given the Perl script works fine for me.

I believe it also runs more efficiently on my machines. I never really took notes on the matter but when I tried the newer approach I had greater system load and being UK based masschecks are ideally run at a time of naturally increased load anyway. It's likely adjustable, but I as there's a method that doesn't need adjusting it was my prefered option.

This is JM's original version and it's what I initially setup my masschecks with, I found it simple and straightforward. The updated method seemed less intuitive to me so when I rebuild masscheck boxes now I prefer to use the original still. The results that get uploaded are still valid data.

Since the code to run this method is still in the repo I would suggest either the fix be applied or the code and documentation be removed.

Btw, I don't know why I show as NoCLA since I do have one. I assume perhaps it was never linked to my bugzilla account? However I would assume the patch as trivial enough for it not to be an issue.

In trunk automasscheck-minimal.sh no more requires bash(1) and is *BSD compatible.

That's not my experience, but regardless, the method I use is included in the official codebase and doesn't work without the suggested patch. However I can see there's no interest in fixing the problem so there's not much point in leaving this open.

IMHO the patch seems fine, it should be committed or we should change the related documentation on the wiki.

Committed attached patch in trunk revision 1820470.