[Openmcl-devel] ccl 1.11 pre-release

[Gary Byers]

If anyone is actually interested in this change and its impact, there is 
a (typo-ridden) discussion of some of that
in http://trac.clozure.com/ccl/ticket/1292.

Much of that discussion took place two months ago.

I find the issues clear - implementations should be free to change 
implementation details, and people who use
libraries that depend on implementation details are taking a risk in 
doing so.

The maintainer of the NAMED-READTABLES library took responsibility for 
that risk.  I shouldn't be, but  I am surprised to see that
what seems like a simple and clear issue is apparently causing confusion 
as long as it has.

//

On 09/09/2015 07:46 PM, Stas Boukarev wrote:
> Anton Vodonosov <avodonosov at yandex.ru> writes:
>
>> 08.09.2015, 22:50, "R. Matthew Emerson" <rme at clozure.com>:
>>> A pre-release version of Clozure CL 1.11 is now available for testing.
>>>
>>> Please see http://ccl.clozure.com/blog/?p=178
>> We tested it with cl-test-grid on linux and macosx.
>>
>> As we can see in the reports below, on both OSes there are regressions. Most of them
>> are caused by the same two (interwinded) errors:
>>
>> 1. Some problem with named-readtables
>> 2. When the readtables error occurs, handling the condition causes some "Undefined function CCL::RDTAB.ALIST called with ..."
>>
>> https://common-lisp.net/project/cl-test-grid/ccl/ccl-1.11-rc1-linux-diff.html
>> https://common-lisp.net/project/cl-test-grid/ccl/ccl-1.11-rc1-macosx-diff.html
>>
>> To reproduce, you can for example (ql:quickload :cl-bson)
> named-readtables was fixed a while ago.
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clozure.com/pipermail/openmcl-devel/attachments/20150910/5c852c83/attachment.htm>