14
$\begingroup$

Bug introduced in 10.2 and persisting through 10.2 or later


Note that the analysis of this problem revealed a different problem than in Potential pollution of Global` context in fresh kernel (see answers). Despite the initial impression that is reflected in the title, the "Global`" context is empty at startup.

enter image description here

Bug again in V10.2?

Fresh kernel.

Linux is OK: enter image description here

$\endgroup$
17
  • 6
    $\begingroup$ Confirmed in v10.2 under Mac OS X 10.10.4. $\endgroup$
    – Cassini
    Commented Aug 1, 2015 at 17:10
  • 2
    $\begingroup$ Out of curiosity I grepped the installation directory and m7res appears only in PLI.mx in SystemFiles/Components/PLI. Does anyone know what PLI stands for? $\endgroup$
    – Szabolcs
    Commented Aug 1, 2015 at 17:20
  • 3
    $\begingroup$ @Szabolcs PLI stands for "Programmable Linguistic Interface". It is in the Semantic` context, which contains things like parsers for various inputs into Dataset objects, although it is not clear to me exactly what it does here or how it works (it seems to be mostly cloud-related). See symbols under Semantic`PLIDump` . $\endgroup$ Commented Aug 1, 2015 at 18:31
  • 4
    $\begingroup$ I just want to respond to the closure and justify my casting of a re-open vote. This is the same type of bug as in the other question, but it is not the same bug. The last one was fixed; this one isn't, and it isn't a regression either, because the symbols are different, as are the circumstances under which the leak appears. $\endgroup$ Commented Aug 2, 2015 at 18:45
  • 3
    $\begingroup$ @Oleksandr Might as well have two questions: one for sym and one for $m7res, since these two leaks happen to be separate bugs, coming from different pieces of code and not introduced at the same time. Each of them can be triggered to appear in Global` independently of the other. One is currently fixed in the development version, the other is being looked into. $\endgroup$
    – ilian
    Commented Aug 2, 2015 at 19:53

1 Answer 1

10
$\begingroup$

I think the symbols sym and $m7res are created by Information. They are not present when the kernel is started.

Fresh kernel 1:

Quit[]

Names["Global`*"]
(*  {}  *)

enter image description here

Fresh kernel 2:

Quit[]

foo = Trace[
   Information["Global`*"],
   TraceInternal -> True];

foo[[8, 3, 5, 7, 2, 9, 18, 65, 2, 1, 3, 6, 7, 5, 6, 4, 3, 3, 6, 2, 8, 
  12, 10, 5, 6, 2, 3, 4, 76]]

foo[[8, 3, 5, 7, 2, 9, 19, 14, 2, 3, 7, 785]]

enter image description here

Warning: If you try turning on General::newsym, you will discover that an overwhelming number of new symbols are created. (You can turn off General::stop if you are really interested in seeing them all.)

$\endgroup$
8
  • 1
    $\begingroup$ Somebody forgot to make his variables private… $\endgroup$ Commented Aug 2, 2015 at 13:57
  • $\begingroup$ @Guesswhoitis. I'm surprised they're still tinkering with Information. $\endgroup$
    – Michael E2
    Commented Aug 2, 2015 at 13:59
  • $\begingroup$ To be fair, the output of ?? in old Mathematica didn't have the fancy formatting. Clearly, they are trying to spruce it up. $\endgroup$ Commented Aug 2, 2015 at 14:02
  • $\begingroup$ @Guesswhoitis. That formatting has been around for a few major releases, at least since V8. It may have changed slightly. I wouldn't be able to remember or notice, probably. $\endgroup$
    – Michael E2
    Commented Aug 2, 2015 at 14:11
  • 2
    $\begingroup$ What an oddity. Good job in narrowing this down. I suppose it also suggests that a huge number of symbols are being created all the time in obscure contexts, most of them having no usefully persistent definitions and never to be used again. I wonder what impact all of this has on performance? $\endgroup$ Commented Aug 2, 2015 at 18:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.