IRC channel logs

2013-05-11.log

back to list of logs

***Guest21872 is now known as eMBee
***eMBee is now known as eM|Bee
***eM|Bee is now known as eMBee
***eMBee is now known as eM|Bee
***eM|Bee is now known as eM-Bee
***eM-Bee is now known as em-bee
***em-bee is now known as embee
***embee is now known as em-bee
***fangism is now known as fangism-shadow
***brendyyn is now known as brendyn
<dsmith>ijp, Why not? (about the macro primer)
<stis>morning guilers!
*stis is fighting with a really difficult bug in guile-log.
<stis>strangely, finding bugs is almost as fun as writng code!
<stis>err /finding/finding and solving/
<taylanub>How should I test floats for equality ?
<alexei>(= 0.0 -0.0)
<taylanub>Nevermind, I asked wrong ..
<taylanub>I'm using `bytevector-ieee-single-native-set!' and `bytevector-ieee-single-native-ref' and the value that comes out isn't the same anymore ..
<taylanub>But that's OK because I'm really writing tests for some code that wraps those procedures, so I can just use the bare procedures in my test and compare the results.
<civodul>in general, floats cannot be compared for equality
<alexei>I guess literal 0.3 would be in doble precision which is lost when you store it into single precision slot
<taylanub>In C there's these constants you use, right ? (= (abs (- float1 float2)) flt_something)
<taylanub>"Epsilon" it was called I think.
<taylanub>So do we have epsilon constants for IEEE floats and doubles ?..
***sneek_ is now known as sneek
***ijp` is now known as ijp
***micro___ is now known as micro_
<dsmith>Hey
<stis>Howdy:
***micro_ is now known as Guest4094
<rlb>ok, so now that the freeze is over, and fwiw, I have 2.0.9 debs about ready to go, but I can't upload them until automake 1.12 moves from experimental to sid.
<stis>yey! Bug is squashed!
<rlb>don't know if this is a known problem, but the build fails for me with "@include: could not find effective-version.texi". I'll poke around when I get back later.
<rlb>