Allegro 8.0 Lisp Also Crashes On Mac OSX Like SBCL Does
John Wiseman sadly reports that the latest Allegro version of their Common Lisp, 8.0, which has gotten written up in eWeek and made us all proud, also crashes (not really) on Mac OSX the same way that SBCL does. This is a problem that’s going to have to be fixed, because the number of CrashReporter instances it generates on a system is unacceptable when in REPL developing.
Imagine every stacktrace spawning a 30MB process on your system each time…
I also wrote something about it as a note (read:warning) on my research project’s installation instructions. It goes:
You can use SBCL on Mac OSX from both the command line and visually, but you probably don’t want to at this moment. SBCL throws a SIGSEGV at every stracktrace, and Apple’s Crash Reporter looks for these events at the kernel level and opens a Crash Reporter process. When being used visually, SBCL will give you a dialogue box displaying this issue. However, when you run SBCL from the command line or even in the REPL in interactive mode (like you do in Emacs with SLIME), every compile will trigger an Apple Crash Reporter process. This means that after one hour of development you could have 30 or 40 crash reporters launched on your system without any visual indication. Since SBCL developers are aware of the problem, but have not patched the SBCL distribution, we do not recommend or support using SBCL in development on Mac OSX. You can still do it, but keep your Activity Viewer open.
Now it would appear the same thing is true with Allegro 8.0, but something tells me this will be fixed quick. There are a lot of scientist running OSX in the biomedical and computer science field and they are Allegro’s bread and butter.
I don’t think we’ll get the same from SBCL.
February 10th, 2006 at 5:30 am
SBCL’s bread and butter is contributions from the user community, to fix bugs that bother them and add features that they need (good quality bug reports are a positive thing, but they’re only bread, not with butter added). A suggestion of where to start was made by Gary Byers in the comments to John Wiseman’s posting about SBCL; please ask if you have any more questions.
Christophe
February 10th, 2006 at 8:02 am
That is true, and I do like SBCL as a Lisp (I actively support it in my work as well as Allegro). My intention in bringing this up is to show that even the big guys got roped in to this bug, so SBCL can’t be blamed as the only distro that has this problem.
I wish we had a larger or at least commercial free full implementation of Common Lisp like Sun has done with Java, leaving the source open but licensed. Half of the weakness of Lisp is that when Allegro does fix this bug, chances are they will not share the fix with the community or document what they did to resolve it. This is different from other languages which are much more open.
I do beleive, however, that this will be changing in the future, and hopefully in the next few years SBCL and/or CLISP will surpass Allego and LispWorks as the primary mindshare of Lisp users, since it is obvious that the large commercial companies do not wish to help fan the flames of this new Lisp resurgance, even though it will benefit them more in the future in increased sales.