Re: [PATCH] Fix for 'make check' failure on non-mule UNIX XEmacs.

Ben> i'm not sure what the issue is. malcolm is right that this test is Ben> simply testing the working of eol detection and obviously won't work when Ben> eol detection is not enabled. Following the discussion as Stephen suggested I came across the problems that prevented it being checked in then: First of all, the coding system no-conversion-coding-system-mapping does not seem to exist (I get errors trying to set it). Is this really the case? Secondly, set-eol-detection changes the eol flag for the coding system buffer-file-coding-system-for-read, but in this test the buffer-coding-system is used (see file-coding.c:3712). I work around this by setting both buffer-file-coding-system-for-read and default-buffer-file-coding-system, but the documentation says that the buffer-coding-system is only used on write but here it is used during a read. I guess that changing eol detection will not affect writes, but does this work around look like the correct action? The issue is the need to add `default-buffer-file-coding-system' to the list that set-eol-detection uses. It seems that, WRT eol-detection, the unit tests are not run in the same environment as normal XEmacs usage (eg this test fails on my machines even if XEmacs is built with --with-default-eol-detection) so there is a question as to whether or not this change to set-eol-detection will not have wider implications. It seems, Ben, that your the only one what can provide a definite answer. Malcolm -- Malcolm Purvis <malcolmp(a)xemacs.org&gt;