[UPDATES] 5 more for 3.9alpha
David T. Lewis
lewis at mail.msen.com
Sat Dec 4 16:15:01 UTC 2004
On Sat, Dec 04, 2004 at 03:59:32PM +0100, Marcus Denker wrote:
> Which of these should be in 3.8?
I don't remember the guidelines exactly, but PrintShowingDecimalPlaces
is a minor fix, so it is not important to put in 3.8. I can't say about
the others.
> 6502PointRectBasicNew-md -- Marcus Denker -- 2 December 2004
> makes creation methods of Point and Rectangle use basicNew instead of
> new, thus
> saving a call to initialize."
>
> 6503DelCompareTiming -- Marcus Denker -- 2 December 2004
> removes #compareTiming form Symbol and Multisymbol:
> Subject: [BUG]Symbol>compareTiming
> Author: Tim Rowledge
> Date Posted: 27 June 2003
> Archive ID: 10815
> Comments: What on earth is this method doing in Symbol? It has nothing
> to do with
> Symbol and if it has any value at all ought to be pulled out and put in
> the benchmarks package.
>
> 6504PrintShowingDecimalPlacesTest-dtl -- David T. Lewis -- 26 November
> 2004
> A unit test for Number>>printShowingDecimalPlaces:
> This will fail for ''10 printShowingDecimalPlaces: 20'' due to missing
> range check in Utilities class>>floatPrecisionForDecimalPlaces:
> A fix is provided in a separate change set."
>
> 6505removeLF-md -- Marcus Denker -- 24 November 2004
> The code run in the postscript is from
> SystemDictionary>>removeAllLineFeeds but
> it does not beep nor ask if there''s a single LF (it just replaces this
> with CR, too)
> This will change over 500 Methods"
>
> 6506PrintShowingDecimalPlacesFix-dtl -- David T. Lewis -- 26 November
> 2004
> Adds a range check for Utilities>>floatPrcisionForDecimalPlaces: and
> a default implementation that works when the parameter is out of range.
> Implementation note:
> I used #caseOf:otherwise: to implement this. The result is about 25%
> slower
> than the original unchecked method. I got very slightly faster
> performance
> by writing it with a bunch of #ifTrue: lines, but I think the caseOf
> approach
> is slightly less ugly.
> All other implementations that I tried (including #at:ifAbsent:, and
> #on:do)
> were significantly slower, so I settled on #caseOf:otherwise: for
> clarity and
> reasonable speed."
>
More information about the Squeak-dev
mailing list
|