I did not remove this method, because I was not sure if this method is
really redundant to Pharo? >From what I understand this test checks if #testConvertToSmalltalkNewlines: can handle unicode characters beyond the byte-range (the euro symbol). Lukas On 16 April 2010 00:14, <[hidden email]> wrote: > John O'Keefe uploaded a new version of Grease-Tests-Pharo-Core to project Seaside 3.0: > http://www.squeaksource.com/Seaside30/Grease-Tests-Pharo-Core-jok.7.mcz > > ==================== Summary ==================== > > Name: Grease-Tests-Pharo-Core-jok.7 > Author: jok > Time: 15 April 2010, 4:13:48 pm > UUID: 427a0773-62f9-f04a-ac30-cbb8db7bc866 > Ancestors: Grease-Tests-Pharo-Core-lr.6 > > remove redundant #testConvertToSmalltalkNewlines > > =============== Diff against Grease-Tests-Pharo-Core-lr.6 =============== > > Item was removed: > - ----- Method: GRPharoPlatformTest>>testConvertToSmalltalkNewlines (in category 'tests') ----- > - testConvertToSmalltalkNewlines > - | expected euro | > - expected := 'selector' , (String with: Character cr) , '"comment"'. > - self assert: expected = (GRPlatform current convertToSmalltalkNewlines: expected). > - self > - assert: > - expected = (GRPlatform current convertToSmalltalkNewlines: 'selector' , (String with: Character lf) , '"comment"'). > - self > - assert: > - expected > - = > - (GRPlatform current > - convertToSmalltalkNewlines: 'selector' , (String with: Character cr with: Character lf) , '"comment"'). > - euro := String with: (Character codePoint: 8364). > - self assert: (expected , euro) = (GRPlatform current convertToSmalltalkNewlines: expected , euro). > - self assert: (GRPlatform current convertToSmalltalkNewlines: nil) isNil! > > > _______________________________________________ > commits mailing list > To unsubscribe, email [hidden email] > http://lists.seaside.st/listinfo/commits > -- Lukas Renggli www.lukas-renggli.ch _______________________________________________ seaside-dev mailing list [hidden email] http://lists.squeakfoundation.org/mailman/listinfo/seaside-dev |
Sorry for the delay in responding -- this note got lost in the
blizzard of GSoC stuff. GreaseCoreTests contains a lot of tests using unicode characters, so why would this one be Pharo-specific? Couldn't/Shouldn't it just be added to GRPlatformTest>>#testConvertToSmalltalkNewlines if there is some concern about newline conversion and unicode behaving badly together? I admit that I didn't even look to see if it was there -- I assumed (badly) that it was. I would be happy to put it there to relieve the concern. John On Thu, Apr 15, 2010 at 4:26 PM, Lukas Renggli <[hidden email]> wrote: > I did not remove this method, because I was not sure if this method is > really redundant to Pharo? > > >From what I understand this test checks if > #testConvertToSmalltalkNewlines: can handle unicode characters beyond > the byte-range (the euro symbol). > > Lukas > > On 16 April 2010 00:14, <[hidden email]> wrote: >> John O'Keefe uploaded a new version of Grease-Tests-Pharo-Core to project Seaside 3.0: >> http://www.squeaksource.com/Seaside30/Grease-Tests-Pharo-Core-jok.7.mcz >> >> ==================== Summary ==================== >> >> Name: Grease-Tests-Pharo-Core-jok.7 >> Author: jok >> Time: 15 April 2010, 4:13:48 pm >> UUID: 427a0773-62f9-f04a-ac30-cbb8db7bc866 >> Ancestors: Grease-Tests-Pharo-Core-lr.6 >> >> remove redundant #testConvertToSmalltalkNewlines >> >> =============== Diff against Grease-Tests-Pharo-Core-lr.6 =============== >> >> Item was removed: >> - ----- Method: GRPharoPlatformTest>>testConvertToSmalltalkNewlines (in category 'tests') ----- >> - testConvertToSmalltalkNewlines >> - | expected euro | >> - expected := 'selector' , (String with: Character cr) , '"comment"'. >> - self assert: expected = (GRPlatform current convertToSmalltalkNewlines: expected). >> - self >> - assert: >> - expected = (GRPlatform current convertToSmalltalkNewlines: 'selector' , (String with: Character lf) , '"comment"'). >> - self >> - assert: >> - expected >> - = >> - (GRPlatform current >> - convertToSmalltalkNewlines: 'selector' , (String with: Character cr with: Character lf) , '"comment"'). >> - euro := String with: (Character codePoint: 8364). >> - self assert: (expected , euro) = (GRPlatform current convertToSmalltalkNewlines: expected , euro). >> - self assert: (GRPlatform current convertToSmalltalkNewlines: nil) isNil! >> >> >> _______________________________________________ >> commits mailing list >> To unsubscribe, email [hidden email] >> http://lists.seaside.st/listinfo/commits >> > > > > -- > Lukas Renggli > www.lukas-renggli.ch > _______________________________________________ > seaside-dev mailing list > [hidden email] > http://lists.squeakfoundation.org/mailman/listinfo/seaside-dev > -- John O'Keefe [|], Principal Smalltalk Architect, Instantiations Inc. _______________________________________________ seaside-dev mailing list [hidden email] http://lists.squeakfoundation.org/mailman/listinfo/seaside-dev |
On 04/19/2010 10:37 PM, John O'Keefe wrote:
> Sorry for the delay in responding -- this note got lost in the > blizzard of GSoC stuff. > > GreaseCoreTests contains a lot of tests using unicode characters, so > why would this one be Pharo-specific? Couldn't/Shouldn't it just be > added to GRPlatformTest>>#testConvertToSmalltalkNewlines if there is > some concern about newline conversion and unicode behaving badly > together? I admit that I didn't even look to see if it was there -- I > assumed (badly) that it was. I would be happy to put it there to > relieve the concern. We did get those tests to pass under GNU Smalltalk; however, this was probably not the best implementation possible, for two reasons. First, introducing a lot of encoding and decoding is going to hurt performance. Second, in GNU Smalltalk you cannot #nextPutAll: a non-encoded String onto an encoded String, so it hurts portability too. So, for Seaside 3.0 (and Iliad 0.8, which is using Grease) we're supporting the null codec only. For 3.1, as already discussed, I'd suggest rethinking the codecs to do triangulation between encodings automatically. This would be much faster for gst because the triangulation is done by the C library via iconv, and would bypass the performance and portability problems above. Paolo _______________________________________________ seaside-dev mailing list [hidden email] http://lists.squeakfoundation.org/mailman/listinfo/seaside-dev |
Free forum by Nabble | Edit this page |