# New Ticket Created by "Carl Mäsak" # Please include the string: [perl #124191] # in the subject line of all future correspondence about this issue. # <URL: https://rt.perl.org/Ticket/Display.html?id=124191 >
<[Tux]> Something recent broke Text::CSV :( <jnthn> [Tux]: How does it fail? * [Tux] is investigating <[Tux]> if I isolate the test, the test passes <[Tux]> If anyone feels like digging: https://github.com/Tux/CSV/blob/master/t/55_combi.t#L46 <= is the line that fails <[Tux]> https://gist.github.com/Tux/7d23f5478dc962699f8f <= REPRODUCED! <[Tux]> please fix :) <[Tux]> memory leak or something? <[Tux]> jnthn, ^^ <jnthn> Doubt it's a leak, could be corruption <jnthn> Either way, a good find <[Tux]> tests++ <jnthn> I'm working on $other-hard-problem at the moment <jnthn> But will put it on my todo list :) <masak> surely that can be golf'd a bit? * masak tries <masak> reproduced locally. <masak> and at exactly the same point in the loop. <masak> seems very stable that way. <masak> [Tux]: I managed to golf it quite a bit. <masak> m: class C is Exception { method message { "OH NOES" } }; for ^208 { { die C.new; CATCH { default {} } }; print "." } <camelia> rakudo-moar fb991b: OUTPUT«...............................................................................................................................................................................................................OH NOES in block [...] <masak> m: my $dots = 0; class C is Exception { method message { "That is $dots dots" } }; for ^208 { { die C.new; CATCH { default {} } }; $dots++ } <camelia> rakudo-moar fb991b: OUTPUT«That is 207 dots in block [...] <masak> so it *works* for 207 iterations, and then -- GC triggers? JIT? optimizer? -- it fails on the 208th. * masak submits rakudobug <arnsholt> Weird! <masak> Tux says something recent caused the above failure. I'd really appreciate it if someone wanted bisect (maybe a month back or so). <[Tux]> last PASS was yesterday or wednesday <[Tux]> I rebuild rakudo daily <jnthn> There was a Moar version bump yesterday, for the first time since the 2015.03 release. <jnthn> -2015.03 <jnthn> +2015.03-16-ga752064 <masak> oh, that's a prime candidate, then. <jnthn> Curiously, the only spesh patches I see in the list change the thresholds for when dynamic opt kicks in. <jnthn> Meaning the problem in theory woulda been noticable *sooner* before...