seaside-dev September 2022

seaside-dev@lists.squeakfoundation.org

8 participants
4 discussions

#substrings:
by Philippe Marschall 06 Sep '22

06 Sep '22

Hello While looking at some of our Squeak 6 test failures I noticed that GRStringTest >> #testSubStrings is testing String >> #substrings:. This is a bit surprising / confusing because: - String >> #subStrings: is ANSI but String >> #substrings: is not. - The method name and comments in #testSubStrings suggest what we actually want to test is #subStrings:, not #substrings:. - The only sender we have of String >> #substrings: is GRStringTest >> #testSubStrings I see the following options: - leave as is - remove the tests and extension methods from GemStone and Squeak - switch to #subStrings: and add it as extension method to Pharo - introduce #greaseSubStrings: Do you have any opinions? Cheers Philippe

7 9

Re: [Seaside-dev] empty collection test failures
by Adriaan van Os 02 Sep '22

02 Sep '22

Hmmm.... VAST does what Pharo/Squeak did before: #(1 2 3) endsWithSubCollection: #(3). true "<- VAST (11)" #(1 2 3) endsWithSubCollection: #(). false "<- VAST (11)" #(1 2 3) beginsWithSubCollection: #(1). true "<- VAST (11)" #(1 2 3) beginsWithSubCollection: #(). false "<- VAST (11)" But what about the following? #() beginsWithSubCollection: #(). false "<- VAST (11)" #() endsWithSubCollection: #(). false "<- VAST (11)" In VAST both #beginsWithSubCollection: and #endsWithSubCollection: use #indexOfSubCollection:startingAt:, which is in an ANSI protocol. The comment in #indexOfSubCollection:startingAt: reads: "[..] Answer 0 if no such subsequence is found in the receiver, or if @targetSequence is empty. [..] Answer 0 if the receiver is empty" Answering 0 by #indexOfSubCollection:startingAt: results in #beginsWithSubCollection: and #endsWithSubCollection: to return false subsequently. What behaviour do we want/need in Seaside? Looks like Seaside has no senders with an empty argument. So maybe not bother and drop the test? Cheers, Adriaan.

2 1

Re: [Seaside-dev] #substrings:
by Adriaan van Os 02 Sep '22

02 Sep '22

FWIW, VAST has EsString>>#subStrings: and UnicodeString>>#subStrings:. I would say: settle for #subStrings: and add it as extension method to Pharo. Cheers, Adriaan.

3 2

empty collection test failures
by Philippe Marschall 01 Sep '22

01 Sep '22

Hello I had a look at our Grease test failures in Pharo 11 and Squeak 6. It seems like on both platforms #endsWith: and #beginsWith now answer true for empty arguments. The comment even says as much. "Some platforms implement #endsWith: to answer true for an empty argument." To me this behavior makes more sense than what we have right now. To give you an illustrative example: #(1 2 3) endsWithSubCollection: #(3). "true" #(1 2 3) endsWithSubCollection: #(). "false in Pharo 10 and Squeak 5, true in Pharo 11 and Squeak 6 " #(1 2 3) beginsWithSubCollection: #(1). "true" #(1 2 3) beginsWithSubCollection: #(). "false in Pharo 10 and Squeak 5, true in Pharo 11 and Squeak 6 " How should we proceed here? - Remove the test for empty collections and leave the behavior unspecified. - Keep the tests as is but change the implementations of #beginsWithSubCollection: and #endsWithSubCollection: to return false on empty collections. - Change the tests as is but change the implementations of #beginsWithSubCollection: and #endsWithSubCollection: to return true on empty collections. Cheers Philippe

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

seaside-dev September 2022