On Apr 10, 2015, at 2:45 PM, Matthew Flatt <[email protected]> wrote:

> At Fri, 10 Apr 2015 13:02:09 -0400, Ryan Culpepper wrote:
>> * Matthew Flatt <[email protected]>
>>   - R6RS Tests
> 
> The build includes the test repair that I made after the last attempt,
> but I overlooked another test with exactly the same problem.
> 
> That mistake was fortunate, in a way, because it made me revisit the
> Unicode specification to make sure that the test should change. I
> discovered that the definition of "case-ignorable" has changed, and
> Racket's update to Unicode 7.0 needs to be adjusted accordingly: an
> apostrophe does not have the a MidLetter value, anymore, but its very
> own Single_Quote value now counts as case-ignorable. It seems that the
> old test was right after all, so I'll fix `string-titlecase` and revert
> the test.
> 
>>   - Run the unix installer tests
> 
> Failed, mainly due to timestamp issues in the build.
> 
> The bytecode files in "collects" for the distributions all have
> timestamps of March 17. Is that intentional? That doesn't cause any
> problems, but it seems strangely out of sync with the date that the
> build must have happened.

There has been only one commit cherry-picked onto the main repository since the 
release branch was created that would have caused the recompilation of any zo 
files, and I usually don’t use the CLEAN flag for intermediate release 
candidates. (Should I?)

> The problem comes from the bytecode file for "scribble/xref.rkt" (in
> the "scribble-lib" package) having a timestamp 12 seconds later than
> the bytecode file for "setup/xref.rkt" (in the "racket-index" package),
> even though "setup/xref.rkt" depends on "scribble/xref.rkt”.

racket-index/setup/scribble.rkt is in fact the subject of that one commit. Is 
it possible that the build process caches built pkgs even when the timestamps 
of their dependencies change? A typical release build involves a “make site” 
followed by one or more “make installers-from-built” to deal with random 
problems (usually on Windows) followed by a “make site-from-installers”.

It’s also possible that I ran a previous build with cherry-picked changes on 
the local release branch but before pushing them to pltgit. Maybe that 
discrepancy caused the problem. I’ll try doing a build from a fresh checkout 
and see if that fixes the problem.

Ryan

-- 
You received this message because you are subscribed to the Google Groups 
"Racket Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/racket-dev/4E03F32A-F561-42FF-87B2-A0E94B2B9A1D%40ccs.neu.edu.
For more options, visit https://groups.google.com/d/optout.

Reply via email to