On 15 Dec 2018, at 15:50, Paul Schmehl wrote:

--On December 14, 2018 at 4:09:08 PM -0500 Bill Cole <postfixlists-070...@billmail.scconsult.com> wrote:

On 14 Dec 2018, at 15:32, Bill Cole wrote:

On 14 Dec 2018, at 0:46, Paul Schmehl wrote:

I'm seeing this error in the logs:

warn: Unescaped left brace in regex is deprecated here (and will be
fata))

NOTE: that message should specify the source of the error. If it does
not, something in your logging plumbing is mangling error messages.

Also note: there is at least one 3rd-party ruleset (http://sa.zmi.at)
which appears to have had a typo ('.' where a ',' should have been) which
also resulted in this error message in a recent version. See
https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7669 for a report of this issue (which is not a bug in SA per se, as the rules with the typo
are an independent distribution.)

I did the following searches trying to locate this issue:

# grep -ro "\[,.\]" /usr/local/lib/perl5/site_perl/*

The warning message only applies to '{' NOT '[' and NOT '('

[...]

# grep -ro "{,.}" /usr/local/lib/perl5/site_perl/*

Try "{,.*}" instead. Also, don't use the 'o' option to grep. It eliminates useful context, specifically whether the pattern is inside a regex.

[...]
The same searches run on SA rules found one possible issue:

# grep -ro "\[,.\]" /usr/local/etc/mail/spamassassin/*
/usr/local/etc/mail/spamassassin/70_sare_specific.cf:[,s

As I said in my prior message: remove that file and any others with 'sare' in their name. They are dangerously obsolete.

Also check /var/db/spamassassin, which is where anything maintained by sa-update will be.

I say possible because in my researching this issue I've seen both curly braces only and also both curly braces and square braces, so I'm not completely certain which (or if both) are a problem.

It's always helpful to look at actual documentation, rather than trust research of secondary sources...

Try 'perldoc perldeprecation' for the definitive truth of deprecation warnings.

At any rate, I don't see any dependency on Text::Unicode, although I wouldn't be surprised if it's used by SA.

It absolutely is not, because Text::Unicode does not exist. :)

SA also does not use Text::Unidecode, which is a last-resort tool for special cases where Unicode characters absolutely must be translated to a reasonable US-ASCII representation.


--
Bill Cole
b...@scconsult.com or billc...@apache.org
(AKA @grumpybozo and many *@billmail.scconsult.com addresses)
Available For Hire: https://linkedin.com/in/billcole

Reply via email to