Fwd: [squeak-dev] The Trunk: Kernel-nice.831.mcz
No time to open a bug, but this should better be fixed before 3.0 release ---------- Forwarded message ---------- From: <commits@source.squeak.org> Date: 2014/1/14 Subject: [squeak-dev] The Trunk: Kernel-nice.831.mcz To: squeak-dev@lists.squeakfoundation.org, packages@lists.squeakfoundation.org Nicolas Cellier uploaded a new version of Kernel to project The Trunk: http://source.squeak.org/trunk/Kernel-nice.831.mcz ==================== Summary ==================== Name: Kernel-nice.831 Author: nice Time: 14 January 2014, 3:43:29.851 am UUID: 5c62dca4-bf8e-4330-9cb1-9cd24047aac6 Ancestors: Kernel-nice.830 1) Fix awfull bug in NumberParser when reading a number with fraction part (the trailing digits of the fraction might be skipped when there's many of them because of incorrect update of lastNonZero position) This made this test fail: self assert: (Float readFrom: (((1<<1075) reciprocal + (1<<1400) reciprocal) printShowingDecimalPlaces: 2000)) = Float fmin. I just wanted to check NumberParser w.r.t. a former bug of dtoa, but it failed for different reasons http://www.exploringbinary.com/gays-strtod-returns-zero-for-inputs-just-abov... 2) Classify some as yet unclassified (or miss-classified) =============== Diff against Kernel-nice.830 =============== Item was changed: + ----- Method: IllegalResumeAttempt>>defaultAction (in category 'handling') ----- - ----- Method: IllegalResumeAttempt>>defaultAction (in category 'comment') ----- defaultAction "No one has handled this error, but now give them a chance to decide how to debug it. If none handle this either then open debugger (see UnhandedError-defaultAction)" UnhandledError signalForException: self! Item was changed: + ----- Method: IllegalResumeAttempt>>isResumable (in category 'handling') ----- - ----- Method: IllegalResumeAttempt>>isResumable (in category 'comment') ----- isResumable ^ false! Item was changed: + ----- Method: InMidstOfFileinNotification>>defaultAction (in category 'handling') ----- - ----- Method: InMidstOfFileinNotification>>defaultAction (in category 'as yet unclassified') ----- defaultAction self resume: false! Item was changed: ----- Method: NumberParser>>nextLargeIntegerBase:nPackets: (in category 'parsing-large int') ----- nextLargeIntegerBase: aRadix nPackets: nPackets "Form a Large integer with incoming digits from sourceStream. Return this integer, or zero if no digits found. Stop reading when no more digits or when nPackets elementary LargeInteger have been encountered. Count the number of digits and the lastNonZero digit and store them in instVar" + | high nDigitsHigh lastNonZeroHigh low nDigitsLow halfPackets | - | high nDigitsHigh low nDigitsLow halfPackets | halfPackets := nPackets bitShift: -1. halfPackets = 0 ifTrue: [^self nextElementaryLargeIntegerBase: aRadix]. high := self nextLargeIntegerBase: aRadix nPackets: halfPackets. high isLarge ifFalse: [^high]. nDigitsHigh := nDigits. + lastNonZeroHigh := lastNonZero. low := self nextLargeIntegerBase: aRadix nPackets: halfPackets. nDigitsLow := nDigits. nDigits := nDigitsHigh + nDigitsLow. + lastNonZero := lastNonZero = 0 + ifTrue: [lastNonZeroHigh] + ifFalse: [lastNonZero + nDigitsHigh]. - lastNonZero = 0 ifFalse: [lastNonZero := lastNonZero + nDigitsHigh]. ^high * (aRadix raisedToInteger: nDigitsLow) + low!
I copy and pasted this mail to the bug tracker (took 1 second): https://pharo.fogbugz.com/f/cases/12642/bug-in-NumberParser-when-reading-a-n... On 14 Jan 2014, at 03:48, Nicolas Cellier <nicolas.cellier.aka.nice@gmail.com> wrote:
No time to open a bug, but this should better be fixed before 3.0 release
---------- Forwarded message ---------- From: <commits@source.squeak.org> Date: 2014/1/14 Subject: [squeak-dev] The Trunk: Kernel-nice.831.mcz To: squeak-dev@lists.squeakfoundation.org, packages@lists.squeakfoundation.org
Nicolas Cellier uploaded a new version of Kernel to project The Trunk: http://source.squeak.org/trunk/Kernel-nice.831.mcz
==================== Summary ====================
Name: Kernel-nice.831 Author: nice Time: 14 January 2014, 3:43:29.851 am UUID: 5c62dca4-bf8e-4330-9cb1-9cd24047aac6 Ancestors: Kernel-nice.830
1) Fix awfull bug in NumberParser when reading a number with fraction part (the trailing digits of the fraction might be skipped when there's many of them because of incorrect update of lastNonZero position)
This made this test fail: self assert: (Float readFrom: (((1<<1075) reciprocal + (1<<1400) reciprocal) printShowingDecimalPlaces: 2000)) = Float fmin.
I just wanted to check NumberParser w.r.t. a former bug of dtoa, but it failed for different reasons http://www.exploringbinary.com/gays-strtod-returns-zero-for-inputs-just-abov...
2) Classify some as yet unclassified (or miss-classified)
=============== Diff against Kernel-nice.830 ===============
Item was changed: + ----- Method: IllegalResumeAttempt>>defaultAction (in category 'handling') ----- - ----- Method: IllegalResumeAttempt>>defaultAction (in category 'comment') ----- defaultAction "No one has handled this error, but now give them a chance to decide how to debug it. If none handle this either then open debugger (see UnhandedError-defaultAction)"
UnhandledError signalForException: self!
Item was changed: + ----- Method: IllegalResumeAttempt>>isResumable (in category 'handling') ----- - ----- Method: IllegalResumeAttempt>>isResumable (in category 'comment') ----- isResumable
^ false!
Item was changed: + ----- Method: InMidstOfFileinNotification>>defaultAction (in category 'handling') ----- - ----- Method: InMidstOfFileinNotification>>defaultAction (in category 'as yet unclassified') ----- defaultAction
self resume: false!
Item was changed: ----- Method: NumberParser>>nextLargeIntegerBase:nPackets: (in category 'parsing-large int') ----- nextLargeIntegerBase: aRadix nPackets: nPackets "Form a Large integer with incoming digits from sourceStream. Return this integer, or zero if no digits found. Stop reading when no more digits or when nPackets elementary LargeInteger have been encountered. Count the number of digits and the lastNonZero digit and store them in instVar"
+ | high nDigitsHigh lastNonZeroHigh low nDigitsLow halfPackets | - | high nDigitsHigh low nDigitsLow halfPackets | halfPackets := nPackets bitShift: -1. halfPackets = 0 ifTrue: [^self nextElementaryLargeIntegerBase: aRadix]. high := self nextLargeIntegerBase: aRadix nPackets: halfPackets. high isLarge ifFalse: [^high]. nDigitsHigh := nDigits. + lastNonZeroHigh := lastNonZero. low := self nextLargeIntegerBase: aRadix nPackets: halfPackets. nDigitsLow := nDigits. nDigits := nDigitsHigh + nDigitsLow. + lastNonZero := lastNonZero = 0 + ifTrue: [lastNonZeroHigh] + ifFalse: [lastNonZero + nDigitsHigh]. - lastNonZero = 0 ifFalse: [lastNonZero := lastNonZero + nDigitsHigh]. ^high * (aRadix raisedToInteger: nDigitsLow) + low!
Thanks! I will try to post a SLICE this evening with a test. Note that between 3 & 4 am, I have excuses for just pressing forward button on my opened gmail page. A posteriori, if logging on fogbugz, searching for right classification of report, entitling, copy-pasting, accepting just take you one second, I'm very happy to just have pressed the forward button, this obviously was the optimal collective efficiency ;) cheers 2014/1/14 Marcus Denker <marcus.denker@inria.fr>
I copy and pasted this mail to the bug tracker (took 1 second):
https://pharo.fogbugz.com/f/cases/12642/bug-in-NumberParser-when-reading-a-n...
On 14 Jan 2014, at 03:48, Nicolas Cellier < nicolas.cellier.aka.nice@gmail.com> wrote:
No time to open a bug, but this should better be fixed before 3.0 release
---------- Forwarded message ---------- From: <commits@source.squeak.org> Date: 2014/1/14 Subject: [squeak-dev] The Trunk: Kernel-nice.831.mcz To: squeak-dev@lists.squeakfoundation.org, packages@lists.squeakfoundation.org
Nicolas Cellier uploaded a new version of Kernel to project The Trunk: http://source.squeak.org/trunk/Kernel-nice.831.mcz
==================== Summary ====================
Name: Kernel-nice.831 Author: nice Time: 14 January 2014, 3:43:29.851 am UUID: 5c62dca4-bf8e-4330-9cb1-9cd24047aac6 Ancestors: Kernel-nice.830
1) Fix awfull bug in NumberParser when reading a number with fraction part (the trailing digits of the fraction might be skipped when there's many of them because of incorrect update of lastNonZero position)
This made this test fail: self assert: (Float readFrom: (((1<<1075) reciprocal + (1<<1400) reciprocal) printShowingDecimalPlaces: 2000)) = Float fmin.
I just wanted to check NumberParser w.r.t. a former bug of dtoa, but it failed for different reasons http://www.exploringbinary.com/gays-strtod-returns-zero-for-inputs-just-abov...
2) Classify some as yet unclassified (or miss-classified)
=============== Diff against Kernel-nice.830 ===============
Item was changed: + ----- Method: IllegalResumeAttempt>>defaultAction (in category 'handling') ----- - ----- Method: IllegalResumeAttempt>>defaultAction (in category 'comment') ----- defaultAction "No one has handled this error, but now give them a chance to decide how to debug it. If none handle this either then open debugger (see UnhandedError-defaultAction)"
UnhandledError signalForException: self!
Item was changed: + ----- Method: IllegalResumeAttempt>>isResumable (in category 'handling') ----- - ----- Method: IllegalResumeAttempt>>isResumable (in category 'comment') ----- isResumable
^ false!
Item was changed: + ----- Method: InMidstOfFileinNotification>>defaultAction (in category 'handling') ----- - ----- Method: InMidstOfFileinNotification>>defaultAction (in category 'as yet unclassified') ----- defaultAction
self resume: false!
Item was changed: ----- Method: NumberParser>>nextLargeIntegerBase:nPackets: (in category 'parsing-large int') ----- nextLargeIntegerBase: aRadix nPackets: nPackets "Form a Large integer with incoming digits from sourceStream. Return this integer, or zero if no digits found. Stop reading when no more digits or when nPackets elementary LargeInteger have been encountered. Count the number of digits and the lastNonZero digit and store them in instVar"
+ | high nDigitsHigh lastNonZeroHigh low nDigitsLow halfPackets | - | high nDigitsHigh low nDigitsLow halfPackets | halfPackets := nPackets bitShift: -1. halfPackets = 0 ifTrue: [^self nextElementaryLargeIntegerBase: aRadix]. high := self nextLargeIntegerBase: aRadix nPackets: halfPackets. high isLarge ifFalse: [^high]. nDigitsHigh := nDigits. + lastNonZeroHigh := lastNonZero. low := self nextLargeIntegerBase: aRadix nPackets: halfPackets. nDigitsLow := nDigits. nDigits := nDigitsHigh + nDigitsLow. + lastNonZero := lastNonZero = 0 + ifTrue: [lastNonZeroHigh] + ifFalse: [lastNonZero + nDigitsHigh]. - lastNonZero = 0 ifFalse: [lastNonZero := lastNonZero + nDigitsHigh]. ^high * (aRadix raisedToInteger: nDigitsLow) + low!
On 14 Jan 2014, at 18:54, Nicolas Cellier <nicolas.cellier.aka.nice@gmail.com> wrote:
Thanks! I will try to post a SLICE this evening with a test.
Note that between 3 & 4 am, I have excuses for just pressing forward button on my opened gmail page. A posteriori, if logging on fogbugz, searching for right classification of report, entitling, copy-pasting, accepting just take you one second, I'm very happy to just have pressed the forward button, this obviously was the optimal collective efficiency ;)
:-) No problem. Marcus
Nicolas Cellier wrote
if logging on fogbugz, searching for right classification of report, entitling, copy-pasting, accepting just take you one second
:) ----- Cheers, Sean -- View this message in context: http://forum.world.st/Fwd-squeak-dev-The-Trunk-Kernel-nice-831-mcz-tp4736525... Sent from the Pharo Smalltalk Developers mailing list archive at Nabble.com.
participants (3)
-
Marcus Denker -
Nicolas Cellier -
Sean P. DeNigris