|
From: | Rik |
Subject: | Re: wstate cleanup |
Date: | Mon, 22 Dec 2014 14:25:40 -0800 |
>> 3) Unnecessary changes to m-files? >> >> The patch had changes to skewness.m, kurtosis.m, and test.m. Was this >> intentional or the result of script search/replace? > >Not intentional. I was planning to check those changes in separately. > >> In skewness.m, for example, the original code is >> >> ## Verify no "divide-by-zero" warnings >> %!test >> %! wstate = warning ("query", "Octave:divide-by-zero"); >> %! warning ("on", "Octave:divide-by-zero"); >> %! unwind_protect >> %! lastwarn (""); # clear last warning >> %! skewness (1); >> %! assert (lastwarn (), ""); >> %! unwind_protect_cleanup >> %! warning (wstate, "Octave:divide-by-zero"); >> %! end_unwind_protect >> >> and the delta is >> >> --- a/scripts/statistics/base/skewness.m >> +++ b/scripts/statistics/base/skewness.m >> @@ -157,7 +157,7 @@ >> %! skewness (1); >> %! assert (lastwarn (), ""); >> %! unwind_protect_cleanup >> -%! warning (wstate, "Octave:divide-by-zero"); >> +%! warning (wstate); >> %! end_unwind_protect >> >> But this doesn't seem right, because wstate will be "on" or "off", not a >> structure of all the preserved values. > >Then there is a different bug because I'm currently seeing this: > >octave:1> warning ("query", "Octave:str-to-num") >ans = > > scalar structure containing the fields: > > identifier = Octave:str-to-num > state = off > >If it is supposed to just return a string then I guess it has been broken for a while since I also see this behavior for 3.8.2. > My bad. The warning function has always returned a structure. The cleanup you propose is definitely correct. --Rik |
[Prev in Thread] | Current Thread | [Next in Thread] |