[Pharo-project] Legitimate syntax for # ... foo
# "helllooo" foo prints #foo is this correct? -- Best regards, Igor Stasenko AKA sig.
# "helllooo" foo
prints #foo
is this correct?
Probably not, but the standard compiler accepts it. So I had to patch the parser of the refactoring engine to accept a whole series of degraded constructs. Below a small collection of tests documenting some of these strange things: testSymbolLiteral | tree | #(('# foo' #foo) ('#"bar"foo' #foo) ('##foo' #foo) ('###foo' #foo) ('#foo:' #foo:) ('#foo::' #'foo::') ('#foo::bar' #'foo::bar') ('#foo::bar:' #'foo::bar:') ('#foo::bar::' #'foo::bar::')) do: [ :pair | tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start = 1. self assert: tree stop = pair first size ] testStatements | tree | #(('' 0 0) ('.' 0 1) ('| bar |' 0 0) ('| bar | .' 0 1) ('| bar | ..' 0 2) ('foo. bar' 2 1) ('foo. bar.' 2 2) ('foo. bar. .' 2 3) ('. foo. bar' 2 2)) do: [ :each | tree := RBParser parseExpression: each first. self assert: tree statements size = each second. self assert: tree periods size = each last ] testNumberParsing | numbers node | numbers := #(('1' 1) ('-1' -1) ('123' 123) ('123' 123) ('-123' -123) ('1.1' 1.1) ('-1.1' -1.1) ('1.23' 1.23) ('-1.23' -1.23) ('1e3' 1e3) ('1d3' 1d3) ('1q3' 1q3) ('-1e3' -1e3) ('1e-3' 1e-3) ('-1e-3' -1e-3) ('2r1e8' 2r1e8) ('-2r1e8' -2r1e8) ('2r1e-8' 2r1e-8) ('-2r1e-8' -2r1e-8) ('0.50s2' 0.50s2) ('0.500s3' 0.500s3) ('0.050s3' 0.050s3)). numbers do: [ :spec | node := RBParser parseExpression: spec first. self assert: node token source = spec first. self assert: node value = spec second ] Lukas -- Lukas Renggli www.lukas-renggli.ch
Actually I didn't have all the latest code loaded in the open image. There is more, hold your breath: testSymbolNumbers #(('#1' 1) ('#12' 12) ('#12.3' 12.3) ('# 1' 1) ('##1' 1) ('#"bar"1' 1)) do: [ :pair | tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start > 1. self assert: tree stop = pair first size ] On 29 April 2010 21:12, Lukas Renggli <renggli@gmail.com> wrote:
# "helllooo" foo
prints #foo
is this correct?
Probably not, but the standard compiler accepts it. So I had to patch the parser of the refactoring engine to accept a whole series of degraded constructs. Below a small collection of tests documenting some of these strange things:
testSymbolLiteral     | tree |     #(('# foo' #foo) ('#"bar"foo' #foo) ('##foo' #foo) ('###foo' #foo) ('#foo:' #foo:) ('#foo::' #'foo::') ('#foo::bar' #'foo::bar') ('#foo::bar:' #'foo::bar:') ('#foo::bar::' #'foo::bar::')) do: [ :pair |         tree := RBParser parseExpression: pair first.         self assert: tree value = pair second.         self assert: tree start = 1.         self assert: tree stop = pair first size ]
testStatements     | tree |     #(('' 0 0) ('.' 0 1) ('| bar |' 0 0) ('| bar | .' 0 1) ('| bar | ..' 0 2) ('foo. bar' 2 1) ('foo. bar.' 2 2) ('foo. bar. .' 2 3) ('. foo. bar' 2 2)) do: [ :each |         tree := RBParser parseExpression: each first.         self assert: tree statements size = each second.         self assert: tree periods size = each last ]
testNumberParsing     | numbers node |     numbers := #(('1' 1) ('-1' -1) ('123' 123) ('123' 123) ('-123' -123) ('1.1' 1.1) ('-1.1' -1.1) ('1.23' 1.23) ('-1.23' -1.23) ('1e3' 1e3) ('1d3' 1d3) ('1q3' 1q3) ('-1e3' -1e3) ('1e-3' 1e-3) ('-1e-3' -1e-3) ('2r1e8'  2r1e8) ('-2r1e8' -2r1e8) ('2r1e-8' 2r1e-8) ('-2r1e-8' -2r1e-8) ('0.50s2' 0.50s2) ('0.500s3' 0.500s3) ('0.050s3' 0.050s3)).     numbers do: [ :spec |         node := RBParser parseExpression: spec first.         self assert: node token source = spec first.         self assert: node value = spec second ]
Lukas
-- Lukas Renggli www.lukas-renggli.ch
-- Lukas Renggli www.lukas-renggli.ch
I don't think any other st dialect would accept that syntax but Squeak. It's just how the parser/scanner works now, but It wouldn't break much code IMO if we stop supporting these weirdnesses. Nicolas 2010/4/29 Lukas Renggli <renggli@gmail.com>:
Actually I didn't have all the latest code loaded in the open image. There is more, hold your breath:
testSymbolNumbers     #(('#1' 1) ('#12' 12) ('#12.3' 12.3) ('# 1' 1) ('##1' 1) ('#"bar"1' 1)) do: [ :pair |         tree := RBParser parseExpression: pair first.         self assert: tree value = pair second.         self assert: tree start > 1.         self assert: tree stop = pair first size ]
On 29 April 2010 21:12, Lukas Renggli <renggli@gmail.com> wrote:
# "helllooo" foo
prints #foo
is this correct?
Probably not, but the standard compiler accepts it. So I had to patch the parser of the refactoring engine to accept a whole series of degraded constructs. Below a small collection of tests documenting some of these strange things:
testSymbolLiteral     | tree |     #(('# foo' #foo) ('#"bar"foo' #foo) ('##foo' #foo) ('###foo' #foo) ('#foo:' #foo:) ('#foo::' #'foo::') ('#foo::bar' #'foo::bar') ('#foo::bar:' #'foo::bar:') ('#foo::bar::' #'foo::bar::')) do: [ :pair |         tree := RBParser parseExpression: pair first.         self assert: tree value = pair second.         self assert: tree start = 1.         self assert: tree stop = pair first size ]
testStatements     | tree |     #(('' 0 0) ('.' 0 1) ('| bar |' 0 0) ('| bar | .' 0 1) ('| bar | ..' 0 2) ('foo. bar' 2 1) ('foo. bar.' 2 2) ('foo. bar. .' 2 3) ('. foo. bar' 2 2)) do: [ :each |         tree := RBParser parseExpression: each first.         self assert: tree statements size = each second.         self assert: tree periods size = each last ]
testNumberParsing     | numbers node |     numbers := #(('1' 1) ('-1' -1) ('123' 123) ('123' 123) ('-123' -123) ('1.1' 1.1) ('-1.1' -1.1) ('1.23' 1.23) ('-1.23' -1.23) ('1e3' 1e3) ('1d3' 1d3) ('1q3' 1q3) ('-1e3' -1e3) ('1e-3' 1e-3) ('-1e-3' -1e-3) ('2r1e8'  2r1e8) ('-2r1e8' -2r1e8) ('2r1e-8' 2r1e-8) ('-2r1e-8' -2r1e-8) ('0.50s2' 0.50s2) ('0.500s3' 0.500s3) ('0.050s3' 0.050s3)).     numbers do: [ :spec |         node := RBParser parseExpression: spec first.         self assert: node token source = spec first.         self assert: node value = spec second ]
Lukas
-- Lukas Renggli www.lukas-renggli.ch
-- Lukas Renggli www.lukas-renggli.ch
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
I would really like to get rid of that. It makes everything looking so brittle. Stef On Apr 29, 2010, at 9:50 PM, Nicolas Cellier wrote:
I don't think any other st dialect would accept that syntax but Squeak. It's just how the parser/scanner works now, but It wouldn't break much code IMO if we stop supporting these weirdnesses.
Nicolas
2010/4/29 Lukas Renggli <renggli@gmail.com>:
Actually I didn't have all the latest code loaded in the open image. There is more, hold your breath:
testSymbolNumbers #(('#1' 1) ('#12' 12) ('#12.3' 12.3) ('# 1' 1) ('##1' 1) ('#"bar"1' 1)) do: [ :pair | tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start > 1. self assert: tree stop = pair first size ]
On 29 April 2010 21:12, Lukas Renggli <renggli@gmail.com> wrote:
# "helllooo" foo
prints #foo
is this correct?
Probably not, but the standard compiler accepts it. So I had to patch the parser of the refactoring engine to accept a whole series of degraded constructs. Below a small collection of tests documenting some of these strange things:
testSymbolLiteral | tree | #(('# foo' #foo) ('#"bar"foo' #foo) ('##foo' #foo) ('###foo' #foo) ('#foo:' #foo:) ('#foo::' #'foo::') ('#foo::bar' #'foo::bar') ('#foo::bar:' #'foo::bar:') ('#foo::bar::' #'foo::bar::')) do: [ :pair | tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start = 1. self assert: tree stop = pair first size ]
testStatements | tree | #(('' 0 0) ('.' 0 1) ('| bar |' 0 0) ('| bar | .' 0 1) ('| bar | ..' 0 2) ('foo. bar' 2 1) ('foo. bar.' 2 2) ('foo. bar. .' 2 3) ('. foo. bar' 2 2)) do: [ :each | tree := RBParser parseExpression: each first. self assert: tree statements size = each second. self assert: tree periods size = each last ]
testNumberParsing | numbers node | numbers := #(('1' 1) ('-1' -1) ('123' 123) ('123' 123) ('-123' -123) ('1.1' 1.1) ('-1.1' -1.1) ('1.23' 1.23) ('-1.23' -1.23) ('1e3' 1e3) ('1d3' 1d3) ('1q3' 1q3) ('-1e3' -1e3) ('1e-3' 1e-3) ('-1e-3' -1e-3) ('2r1e8' 2r1e8) ('-2r1e8' -2r1e8) ('2r1e-8' 2r1e-8) ('-2r1e-8' -2r1e-8) ('0.50s2' 0.50s2) ('0.500s3' 0.500s3) ('0.050s3' 0.050s3)). numbers do: [ :spec | node := RBParser parseExpression: spec first. self assert: node token source = spec first. self assert: node value = spec second ]
Lukas
-- Lukas Renggli www.lukas-renggli.ch
-- Lukas Renggli www.lukas-renggli.ch
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
On Thu, Apr 29, 2010 at 12:50 PM, Nicolas Cellier < nicolas.cellier.aka.nice@gmail.com> wrote:
I don't think any other st dialect would accept that syntax but Squeak.
VisualWorks accepts it just fine. # "hello" foo => #foo # "hello" (foo # "hello" foo) => #(#foo #foo) I think this is the way the language works. Relax. Celebrate its weirdness. Don't sweat the petty stuff, pet the sweaty stuff: http://www.popsci.com/gadgets/article/2010-04/exclusive-making-elements-one-... It's just how the parser/scanner works now, but It wouldn't break much
code IMO if we stop supporting these weirdnesses.
Nicolas
2010/4/29 Lukas Renggli <renggli@gmail.com>:
Actually I didn't have all the latest code loaded in the open image. There is more, hold your breath:
testSymbolNumbers #(('#1' 1) ('#12' 12) ('#12.3' 12.3) ('# 1' 1) ('##1' 1) ('#"bar"1' 1)) do: [ :pair | tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start > 1. self assert: tree stop = pair first size ]
On 29 April 2010 21:12, Lukas Renggli <renggli@gmail.com> wrote:
# "helllooo" foo
prints #foo
is this correct?
Probably not, but the standard compiler accepts it. So I had to patch the parser of the refactoring engine to accept a whole series of degraded constructs. Below a small collection of tests documenting some of these strange things:
testSymbolLiteral | tree | #(('# foo' #foo) ('#"bar"foo' #foo) ('##foo' #foo) ('###foo' #foo) ('#foo:' #foo:) ('#foo::' #'foo::') ('#foo::bar' #'foo::bar') ('#foo::bar:' #'foo::bar:') ('#foo::bar::' #'foo::bar::')) do: [ :pair | tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start = 1. self assert: tree stop = pair first size ]
testStatements | tree | #(('' 0 0) ('.' 0 1) ('| bar |' 0 0) ('| bar | .' 0 1) ('| bar | ..' 0 2) ('foo. bar' 2 1) ('foo. bar.' 2 2) ('foo. bar. .' 2 3) ('. foo. bar' 2 2)) do: [ :each | tree := RBParser parseExpression: each first. self assert: tree statements size = each second. self assert: tree periods size = each last ]
testNumberParsing | numbers node | numbers := #(('1' 1) ('-1' -1) ('123' 123) ('123' 123) ('-123' -123) ('1.1' 1.1) ('-1.1' -1.1) ('1.23' 1.23) ('-1.23' -1.23) ('1e3' 1e3) ('1d3' 1d3) ('1q3' 1q3) ('-1e3' -1e3) ('1e-3' 1e-3) ('-1e-3' -1e-3) ('2r1e8' 2r1e8) ('-2r1e8' -2r1e8) ('2r1e-8' 2r1e-8) ('-2r1e-8' -2r1e-8) ('0.50s2' 0.50s2) ('0.500s3' 0.500s3) ('0.050s3' 0.050s3)). numbers do: [ :spec | node := RBParser parseExpression: spec first. self assert: node token source = spec first. self assert: node value = spec second ]
Lukas
-- Lukas Renggli www.lukas-renggli.ch
-- Lukas Renggli www.lukas-renggli.ch
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
2010/4/29 Eliot Miranda <eliot.miranda@gmail.com>:
On Thu, Apr 29, 2010 at 12:50 PM, Nicolas Cellier <nicolas.cellier.aka.nice@gmail.com> wrote:
I don't think any other st dialect would accept that syntax but Squeak.
VisualWorks accepts it just fine. # "hello" foo => #foo # "hello" (foo # "hello" foo) => #(#foo #foo) I think this is the way the language works. Â Relax. Â Celebrate its weirdness. Â Don't sweat the petty stuff, pet the sweaty stuff: http://www.popsci.com/gadgets/article/2010-04/exclusive-making-elements-one-...
Maybe I should relax about comment anywhere, but these: [Compiler evaluate: '#( # 2 # -2 )'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '# -2'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '#-2'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '#$a'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '##(1##)1)'] on: Error do: [:exc | exc return: nil]. would all return nil in VW, not so in Squeak. Also, I mind the difference of first two in Squeak. That's hardly predictable and much too weird to my own taste... Nicolas
It's just how the parser/scanner works now, but It wouldn't break much code IMO if we stop supporting these weirdnesses.
Nicolas
2010/4/29 Lukas Renggli <renggli@gmail.com>:
Actually I didn't have all the latest code loaded in the open image. There is more, hold your breath:
testSymbolNumbers     #(('#1' 1) ('#12' 12) ('#12.3' 12.3) ('# 1' 1) ('##1' 1) ('#"bar"1' 1)) do: [ :pair |         tree := RBParser parseExpression: pair first.         self assert: tree value = pair second.         self assert: tree start > 1.         self assert: tree stop = pair first size ]
On 29 April 2010 21:12, Lukas Renggli <renggli@gmail.com> wrote:
# "helllooo" foo
prints #foo
is this correct?
Probably not, but the standard compiler accepts it. So I had to patch the parser of the refactoring engine to accept a whole series of degraded constructs. Below a small collection of tests documenting some of these strange things:
testSymbolLiteral     | tree |     #(('# foo' #foo) ('#"bar"foo' #foo) ('##foo' #foo) ('###foo' #foo) ('#foo:' #foo:) ('#foo::' #'foo::') ('#foo::bar' #'foo::bar') ('#foo::bar:' #'foo::bar:') ('#foo::bar::' #'foo::bar::')) do: [ :pair |         tree := RBParser parseExpression: pair first.         self assert: tree value = pair second.         self assert: tree start = 1.         self assert: tree stop = pair first size ]
testStatements     | tree |     #(('' 0 0) ('.' 0 1) ('| bar |' 0 0) ('| bar | .' 0 1) ('| bar | ..' 0 2) ('foo. bar' 2 1) ('foo. bar.' 2 2) ('foo. bar. .' 2 3) ('. foo. bar' 2 2)) do: [ :each |         tree := RBParser parseExpression: each first.         self assert: tree statements size = each second.         self assert: tree periods size = each last ]
testNumberParsing     | numbers node |     numbers := #(('1' 1) ('-1' -1) ('123' 123) ('123' 123) ('-123' -123) ('1.1' 1.1) ('-1.1' -1.1) ('1.23' 1.23) ('-1.23' -1.23) ('1e3' 1e3) ('1d3' 1d3) ('1q3' 1q3) ('-1e3' -1e3) ('1e-3' 1e-3) ('-1e-3' -1e-3) ('2r1e8'  2r1e8) ('-2r1e8' -2r1e8) ('2r1e-8' 2r1e-8) ('-2r1e-8' -2r1e-8) ('0.50s2' 0.50s2) ('0.500s3' 0.500s3) ('0.050s3' 0.050s3)).     numbers do: [ :spec |         node := RBParser parseExpression: spec first.         self assert: node token source = spec first.         self assert: node value = spec second ]
Lukas
-- Lukas Renggli www.lukas-renggli.ch
-- Lukas Renggli www.lukas-renggli.ch
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
On Thu, Apr 29, 2010 at 1:25 PM, Nicolas Cellier < nicolas.cellier.aka.nice@gmail.com> wrote:
2010/4/29 Eliot Miranda <eliot.miranda@gmail.com>:
On Thu, Apr 29, 2010 at 12:50 PM, Nicolas Cellier <nicolas.cellier.aka.nice@gmail.com> wrote:
I don't think any other st dialect would accept that syntax but Squeak.
VisualWorks accepts it just fine. # "hello" foo => #foo # "hello" (foo # "hello" foo) => #(#foo #foo) I think this is the way the language works. Relax. Celebrate its weirdness. Don't sweat the petty stuff, pet the sweaty stuff:
http://www.popsci.com/gadgets/article/2010-04/exclusive-making-elements-one-...
Maybe I should relax about comment anywhere, but these:
[Compiler evaluate: '#( # 2 # -2 )'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '# -2'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '#-2'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '#$a'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '##(1##)1)'] on: Error do: [:exc | exc return: nil].
would all return nil in VW, not so in Squeak. Also, I mind the difference of first two in Squeak. That's hardly predictable and much too weird to my own taste...
That's fair enough, but that's a different issue to there being allowable whitespace between the literal hash and the literal. In any case, comparing against VW is probably a sensible thing to do. best Eliot
Nicolas
It's just how the parser/scanner works now, but It wouldn't break much code IMO if we stop supporting these weirdnesses.
Nicolas
2010/4/29 Lukas Renggli <renggli@gmail.com>:
Actually I didn't have all the latest code loaded in the open image. There is more, hold your breath:
testSymbolNumbers #(('#1' 1) ('#12' 12) ('#12.3' 12.3) ('# 1' 1) ('##1' 1) ('#"bar"1' 1)) do: [ :pair | tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start > 1. self assert: tree stop = pair first size ]
On 29 April 2010 21:12, Lukas Renggli <renggli@gmail.com> wrote:
# "helllooo" foo
prints #foo
is this correct?
Probably not, but the standard compiler accepts it. So I had to patch the parser of the refactoring engine to accept a whole series of degraded constructs. Below a small collection of tests documenting some of these strange things:
testSymbolLiteral | tree | #(('# foo' #foo) ('#"bar"foo' #foo) ('##foo' #foo) ('###foo' #foo) ('#foo:' #foo:) ('#foo::' #'foo::') ('#foo::bar' #'foo::bar') ('#foo::bar:' #'foo::bar:') ('#foo::bar::' #'foo::bar::')) do: [
:pair
| tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start = 1. self assert: tree stop = pair first size ]
testStatements | tree | #(('' 0 0) ('.' 0 1) ('| bar |' 0 0) ('| bar | .' 0 1) ('| bar | ..' 0 2) ('foo. bar' 2 1) ('foo. bar.' 2 2) ('foo. bar. .' 2 3) ('. foo. bar' 2 2)) do: [ :each | tree := RBParser parseExpression: each first. self assert: tree statements size = each second. self assert: tree periods size = each last ]
testNumberParsing | numbers node | numbers := #(('1' 1) ('-1' -1) ('123' 123) ('123' 123) ('-123' -123) ('1.1' 1.1) ('-1.1' -1.1) ('1.23' 1.23) ('-1.23' -1.23) ('1e3' 1e3) ('1d3' 1d3) ('1q3' 1q3) ('-1e3' -1e3) ('1e-3' 1e-3) ('-1e-3' -1e-3) ('2r1e8' 2r1e8) ('-2r1e8' -2r1e8) ('2r1e-8' 2r1e-8) ('-2r1e-8' -2r1e-8) ('0.50s2' 0.50s2) ('0.500s3' 0.500s3) ('0.050s3' 0.050s3)). numbers do: [ :spec | node := RBParser parseExpression: spec first. self assert: node token source = spec first. self assert: node value = spec second ]
Lukas
-- Lukas Renggli www.lukas-renggli.ch
-- Lukas Renggli www.lukas-renggli.ch
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
2010/4/29 Eliot Miranda <eliot.miranda@gmail.com>:
On Thu, Apr 29, 2010 at 1:25 PM, Nicolas Cellier <nicolas.cellier.aka.nice@gmail.com> wrote:
2010/4/29 Eliot Miranda <eliot.miranda@gmail.com>:
On Thu, Apr 29, 2010 at 12:50 PM, Nicolas Cellier <nicolas.cellier.aka.nice@gmail.com> wrote:
I don't think any other st dialect would accept that syntax but Squeak.
VisualWorks accepts it just fine. # "hello" foo => #foo # "hello" (foo # "hello" foo) => #(#foo #foo) I think this is the way the language works. Â Relax. Â Celebrate its weirdness. Â Don't sweat the petty stuff, pet the sweaty stuff:
http://www.popsci.com/gadgets/article/2010-04/exclusive-making-elements-one-...
Maybe I should relax about comment anywhere, but these:
[Compiler evaluate: '#( # 2 # -2 )'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '# -2'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '#-2'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '#$a'] on: Error do: [:exc | exc return: nil]. [Compiler evaluate: '##(1##)1)'] on: Error do: [:exc | exc return: nil].
would all return nil in VW, not so in Squeak. Also, I mind the difference of first two in Squeak. That's hardly predictable and much too weird to my own taste...
That's fair enough, but that's a different issue to there being allowable whitespace between the literal hash and the literal. Â In any case, comparing against VW is probably a sensible thing to do. best Eliot
Here we go: fix at http://code.google.com/p/pharo/issues/detail?id=2371 Some tests from Lukas would be appreciated :) Nicolas
Nicolas
It's just how the parser/scanner works now, but It wouldn't break much code IMO if we stop supporting these weirdnesses.
Nicolas
2010/4/29 Lukas Renggli <renggli@gmail.com>:
Actually I didn't have all the latest code loaded in the open image. There is more, hold your breath:
testSymbolNumbers     #(('#1' 1) ('#12' 12) ('#12.3' 12.3) ('# 1' 1) ('##1' 1) ('#"bar"1' 1)) do: [ :pair |         tree := RBParser parseExpression: pair first.         self assert: tree value = pair second.         self assert: tree start > 1.         self assert: tree stop = pair first size ]
On 29 April 2010 21:12, Lukas Renggli <renggli@gmail.com> wrote:
# "helllooo" foo
prints #foo
is this correct?
Probably not, but the standard compiler accepts it. So I had to patch the parser of the refactoring engine to accept a whole series of degraded constructs. Below a small collection of tests documenting some of these strange things:
testSymbolLiteral     | tree |     #(('# foo' #foo) ('#"bar"foo' #foo) ('##foo' #foo) ('###foo' #foo) ('#foo:' #foo:) ('#foo::' #'foo::') ('#foo::bar' #'foo::bar') ('#foo::bar:' #'foo::bar:') ('#foo::bar::' #'foo::bar::')) do: [ :pair |         tree := RBParser parseExpression: pair first.         self assert: tree value = pair second.         self assert: tree start = 1.         self assert: tree stop = pair first size ]
testStatements     | tree |     #(('' 0 0) ('.' 0 1) ('| bar |' 0 0) ('| bar | .' 0 1) ('| bar | ..' 0 2) ('foo. bar' 2 1) ('foo. bar.' 2 2) ('foo. bar. .' 2 3) ('. foo. bar' 2 2)) do: [ :each |         tree := RBParser parseExpression: each first.         self assert: tree statements size = each second.         self assert: tree periods size = each last ]
testNumberParsing     | numbers node |     numbers := #(('1' 1) ('-1' -1) ('123' 123) ('123' 123) ('-123' -123) ('1.1' 1.1) ('-1.1' -1.1) ('1.23' 1.23) ('-1.23' -1.23) ('1e3' 1e3) ('1d3' 1d3) ('1q3' 1q3) ('-1e3' -1e3) ('1e-3' 1e-3) ('-1e-3' -1e-3) ('2r1e8'  2r1e8) ('-2r1e8' -2r1e8) ('2r1e-8' 2r1e-8) ('-2r1e-8' -2r1e-8) ('0.50s2' 0.50s2) ('0.500s3' 0.500s3) ('0.050s3' 0.050s3)).     numbers do: [ :spec |         node := RBParser parseExpression: spec first.         self assert: node token source = spec first.         self assert: node value = spec second ]
Lukas
-- Lukas Renggli www.lukas-renggli.ch
-- Lukas Renggli www.lukas-renggli.ch
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
THANKS!!! I'm sure that lukas is occupy to clean his house/repaint the wall/read all the papers he forgot instead of writing his phd that he will help :D with great enthusiasm.
That's fair enough, but that's a different issue to there being allowable whitespace between the literal hash and the literal. In any case, comparing against VW is probably a sensible thing to do. best Eliot
Here we go: fix at http://code.google.com/p/pharo/issues/detail?id=2371
Some tests from Lukas would be appreciated :)
Eliot cleaning some left over or glitches is always good. I think that this is what nicolas wanted to say. like ###f -> #f we could use that for another syntactic extension. #12 -> 12 is not really nice. or #foo::bar: Stef
VisualWorks accepts it just fine.
# "hello" foo => #foo
# "hello" (foo # "hello" foo) => #(#foo #foo)
It's just how the parser/scanner works now, but It wouldn't break much code IMO if we stop supporting these weirdnesses.
Nicolas
2010/4/29 Lukas Renggli <renggli@gmail.com>:
Actually I didn't have all the latest code loaded in the open image. There is more, hold your breath:
testSymbolNumbers #(('#1' 1) ('#12' 12) ('#12.3' 12.3) ('# 1' 1) ('##1' 1) ('#"bar"1' 1)) do: [ :pair | tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start > 1. self assert: tree stop = pair first size ]
On 29 April 2010 21:12, Lukas Renggli <renggli@gmail.com> wrote:
# "helllooo" foo
prints #foo
is this correct?
Probably not, but the standard compiler accepts it. So I had to patch the parser of the refactoring engine to accept a whole series of degraded constructs. Below a small collection of tests documenting some of these strange things:
testSymbolLiteral | tree | #(('# foo' #foo) ('#"bar"foo' #foo) ('##foo' #foo) ('###foo' #foo) ('#foo:' #foo:) ('#foo::' #'foo::') ('#foo::bar' #'foo::bar') ('#foo::bar:' #'foo::bar:') ('#foo::bar::' #'foo::bar::')) do: [ :pair | tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start = 1. self assert: tree stop = pair first size ]
testStatements | tree | #(('' 0 0) ('.' 0 1) ('| bar |' 0 0) ('| bar | .' 0 1) ('| bar | ..' 0 2) ('foo. bar' 2 1) ('foo. bar.' 2 2) ('foo. bar. .' 2 3) ('. foo. bar' 2 2)) do: [ :each | tree := RBParser parseExpression: each first. self assert: tree statements size = each second. self assert: tree periods size = each last ]
testNumberParsing | numbers node | numbers := #(('1' 1) ('-1' -1) ('123' 123) ('123' 123) ('-123' -123) ('1.1' 1.1) ('-1.1' -1.1) ('1.23' 1.23) ('-1.23' -1.23) ('1e3' 1e3) ('1d3' 1d3) ('1q3' 1q3) ('-1e3' -1e3) ('1e-3' 1e-3) ('-1e-3' -1e-3) ('2r1e8' 2r1e8) ('-2r1e8' -2r1e8) ('2r1e-8' 2r1e-8) ('-2r1e-8' -2r1e-8) ('0.50s2' 0.50s2) ('0.500s3' 0.500s3) ('0.050s3' 0.050s3)). numbers do: [ :spec | node := RBParser parseExpression: spec first. self assert: node token source = spec first. self assert: node value = spec second ]
Lukas
-- Lukas Renggli www.lukas-renggli.ch
-- Lukas Renggli www.lukas-renggli.ch
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
arghhhhhhhhhhhhhhhhhhhhhhh may be one day we should clean that too.
testSymbolLiteral | tree | #(('# foo' #foo) ('#"bar"foo' #foo) ('##foo' #foo) ('###foo' #foo) ('#foo:' #foo:) ('#foo::' #'foo::') ('#foo::bar' #'foo::bar') ('#foo::bar:' #'foo::bar:') ('#foo::bar::' #'foo::bar::')) do: [ :pair | tree := RBParser parseExpression: pair first. self assert: tree value = pair second. self assert: tree start = 1. self assert: tree stop = pair first size ]
the string and symbols mixture is not good. #'hkjh 3' is ok and all the rest looks ugly to me.
testStatements | tree | #(('' 0 0) ('.' 0 1) ('| bar |' 0 0) ('| bar | .' 0 1) ('| bar | ..' 0 2) ('foo. bar' 2 1) ('foo. bar.' 2 2) ('foo. bar. .' 2 3) ('. foo. bar' 2 2)) do: [ :each | tree := RBParser parseExpression: each first. self assert: tree statements size = each second. self assert: tree periods size = each last ]
testNumberParsing | numbers node | numbers := #(('1' 1) ('-1' -1) ('123' 123) ('123' 123) ('-123' -123) ('1.1' 1.1) ('-1.1' -1.1) ('1.23' 1.23) ('-1.23' -1.23) ('1e3' 1e3) ('1d3' 1d3) ('1q3' 1q3) ('-1e3' -1e3) ('1e-3' 1e-3) ('-1e-3' -1e-3) ('2r1e8' 2r1e8) ('-2r1e8' -2r1e8) ('2r1e-8' 2r1e-8) ('-2r1e-8' -2r1e-8) ('0.50s2' 0.50s2) ('0.500s3' 0.500s3) ('0.050s3' 0.050s3)). numbers do: [ :spec | node := RBParser parseExpression: spec first. self assert: node token source = spec first. self assert: node value = spec second ]
Lukas
-- Lukas Renggli www.lukas-renggli.ch
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
participants (5)
-
Eliot Miranda -
Igor Stasenko -
Lukas Renggli -
Nicolas Cellier -
Stéphane Ducasse