http://qs1969.pair.com?node_id=1176910


in reply to Re: split problem
in thread split problem

a git bisect shows the first bad commit is 4ecee209 (the first release that contains this commit appears to be v5.21.5)
The bisect is misleading; the issue is with split when it is optimised to split directly to an array (rather than just returning a list which later gets assigned to something). When exactly this optimisation is enabled has changed between releases, and that commit is one of those changes.

The real issue was introduced in perl 5.19.4, and I've just fixed in it in blead with this commit:

commit 71ca73e5fa9639ac33e9f2e74cd0c32288a5040d Author: David Mitchell <davem@iabyn.com> AuthorDate: Wed Nov 30 08:59:01 2016 +0000 Commit: David Mitchell <davem@iabyn.com> CommitDate: Wed Nov 30 09:11:25 2016 +0000 split was leaving PL_sv_undef in unused ary slots This: @a = split(/-/,"-"); $a[1] = undef; $a[0] = 0; was giving Modification of a read-only value attempted at foo line 3. This is because: 1) unused slots in AvARRAY between AvFILL and AvMAX should always +be null; av_clear(), av_extend() etc do this; while av_store(), if st +oring to a slot N somewhere between AvFILL and AvMAX, doesn't bother to +clear between (AvFILL+1)..(N-1) on the assumption that everyone else pla +ys nicely. 2) pp_split() when splitting directly to an array, sometimes over- +splits and has to null out the excess elements; 3) Since perl 5.19.4, unused AV slots are now marked with NULL rat +her than &PL_sv_undef; 4) pp_split was still using &PL_sv_undef; The fault was with (4), and is easily fixed.

Dave.

Replies are listed 'Best First'.
Re^3: split problem
by haukex (Archbishop) on Nov 30, 2016 at 18:56 UTC

    Hi Dave,

    The real issue was introduced in perl 5.19.4, and I've just fixed in it in blead with this commit

    Thank you very much!

    Regards,
    -- Hauke D

Re^3: split problem
by stevieb (Canon) on Dec 01, 2016 at 14:56 UTC

    I just came back to this issue this morning to follow up to see if a bug ticket was opened, and if not, was going to open one.

    Thanks Dave for getting this fix in!