Re: [Factor-talk] when-empty
It is because when-zero preserves the value on the stack if it isn't zero. I think it is easier to see the mistake if you forego using the when-* combinators: : check0 ( n -- ) 0 = [ call-fail ] [ ] if ; : check0 ( n -- ) [ call-fail ] [ ] if-zero ; Both quotations supplied to if-zero has effect ( -- ), but that is not right. The second quotation should lift one extra item from the stack. As in: : check0 ( n -- ) [ call-fail ] [ drop ] if-zero ; When I write Factor code, I always start with the "easy" branch combinators like `if` and `?`. Then after I have ensured that the code does what I want and that the stack effects balance, I look for opportunities to refactor using "fancier" combinators like `when`, `?if`, `unless-empty` and so on. 2017-03-01 20:18 GMT+01:00 Alexander Ilin : > Continuing with this example, here's what I stumbled upon (again), and found > confusing (once more): > > This works as expected: > > ``` > : check0 ( n -- ) 0 = [ call-fail ] when ; > ``` > > This fails to compile: > > ``` > : check0 ( n -- ) [ call-fail ] when-zero ; > ``` > > But according to the documented stack effect (at least the way I read it, > please correct me if I'm wrong) it should compile: n and quotation should be > dropped from the stack by when-zero. > > 01.03.2017, 22:13, "Alexander Ilin" : > > Here's a slightly better stack effect documentation: > > `when-zero ( ..a n quot: ( ..a -- ..b ) -- ..b )` > > But IIUC the actual stack effect is one of these: > > `when-zero ( ..a 0 quot: ( ..a -- ..b ) -- ..b )` > `when-zero ( ..a n#0 quot: ( ..a -- ..b ) -- ..a n )` > > where # denotes inequality. > > Am I right? How does the compiler handle such ambiguity in the stack effect? > 01.03.2017, 19:28, "John Benediktsson" : > > A stack effect of ( seq quot -- ) doesn't say what the quot is expected to > do... > > You can play around with it and see how it works: > > IN: scratchpad "abc" [ "empty" ] when-empty . > "abc" > > IN: scratchpad "" [ "empty" ] when-empty . > "empty" > > We should improve the docs here because they are a little hard to follow, > since it refers to ``when-empty`` taking the "first quot" of an > ``if-empty``, with an empty second quot. > > > > > On Wed, Mar 1, 2017 at 8:18 AM, Alexander Ilin wrote: > > Hello! > > I'm struggling with some of the Factor's words. Mostly with `when-empty`, > `if-zero` and similar. It's difficult for me to understand why they don't > work sometimes, while their simpler counterparts (while, if) do work exactly > as expected. > > Here's the latest installment of my struggle: > > IN: scratchpad USE: crypto.xor [ [ empty-xor-key ] when-empty ] infer. > ( x -- x ) > > Why is the stack effect as shown here? The documentation for `when-empty` > says it takes `( seq quot -- )`. Therefore, I expect `[ [ empty-xor-key ] > when-empty ] infer.` to output `( x -- )`. > > When I modify the quotation passed to `when-empty` to be a no-op, I get a > compilation error: > IN: scratchpad [ [ 1 drop ] when-empty ] infer. > Error > The input quotations to “if-empty” don't match their expected effects > Input Expected Got > [ 1 drop ] ( ..a -- ..b ) ( -- ) > [ ]( ..a seq -- ..b ) ( -- ) > > That's completely weird to me. Is the stack effect in the documentation > incorrect? Should it be `( seq quot -- seq )`? > And why does it refuse to compile the second example with `1 drop`? > > ---=--- > Александр > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > ___ > Factor-talk mailing list > Factor-talk@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/factor-talk > > , > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > > , > > ___ > Factor-talk mailing list > Factor-talk@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/factor-talk > > > > ---=--- > Александр > > , > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > > , > > ___ > Factor-talk mailing list > Factor-talk@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/factor-talk > > > > ---=--- > Александр > > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > ___ > Factor-talk mailing list > Factor-talk@lists
Re: [Factor-talk] when-empty
Continuing with this example, here's what I stumbled upon (again), and found confusing (once more): This works as expected: ```: check0 ( n -- ) 0 = [ call-fail ] when ;``` This fails to compile: ```: check0 ( n -- ) [ call-fail ] when-zero ;``` But according to the documented stack effect (at least the way I read it, please correct me if I'm wrong) it should compile: n and quotation should be dropped from the stack by when-zero. 01.03.2017, 22:13, "Alexander Ilin" :Here's a slightly better stack effect documentation: `when-zero ( ..a n quot: ( ..a -- ..b ) -- ..b )` But IIUC the actual stack effect is one of these: `when-zero ( ..a 0 quot: ( ..a -- ..b ) -- ..b )``when-zero ( ..a n#0 quot: ( ..a -- ..b ) -- ..a n )` where # denotes inequality. Am I right? How does the compiler handle such ambiguity in the stack effect?01.03.2017, 19:28, "John Benediktsson":A stack effect of ( seq quot -- ) doesn't say what the quot is expected to do... You can play around with it and see how it works: IN: scratchpad "abc" [ "empty" ] when-empty . "abc" IN: scratchpad "" [ "empty" ] when-empty . "empty" We should improve the docs here because they are a little hard to follow, since it refers to ``when-empty`` taking the "first quot" of an ``if-empty``, with an empty second quot.On Wed, Mar 1, 2017 at 8:18 AM, Alexander Ilin wrote:Hello! I'm struggling with some of the Factor's words. Mostly with `when-empty`, `if-zero` and similar. It's difficult for me to understand why they don't work sometimes, while their simpler counterparts (while, if) do work exactly as expected. Here's the latest installment of my struggle:IN: scratchpad USE: crypto.xor [ [ empty-xor-key ] when-empty ] infer.( x -- x ) Why is the stack effect as shown here? The documentation for `when-empty` says it takes `( seq quot -- )`. Therefore, I expect `[ [ empty-xor-key ] when-empty ] infer.` to output `( x -- )`. When I modify the quotation passed to `when-empty` to be a no-op, I get a compilation error:IN: scratchpad [ [ 1 drop ] when-empty ] infer.ErrorThe input quotations to “if-empty” don't match their expected effectsInput Expected Got[ 1 drop ] ( ..a -- ..b ) ( -- )[ ] ( ..a seq -- ..b ) ( -- ) That's completely weird to me. Is the stack effect in the documentation incorrect? Should it be `( seq quot -- seq )`? And why does it refuse to compile the second example with `1 drop`?---=--- Александр--Check out the vibrant tech community on one of the world's mostengaging tech sites, SlashDot.org! http://sdm.link/slashdot___Factor-talk mailing listFactor-talk@lists.sourceforge.nethttps://lists.sourceforge.net/lists/listinfo/factor-talk,--Check out the vibrant tech community on one of the world's mostengaging tech sites, SlashDot.org! http://sdm.link/slashdot,___Factor-talk mailing listFactor-talk@lists.sourceforge.nethttps://lists.sourceforge.net/lists/listinfo/factor-talk ---=---Александр ,--Check out the vibrant tech community on one of the world's mostengaging tech sites, SlashDot.org! http://sdm.link/slashdot,___Factor-talk mailing listFactor-talk@lists.sourceforge.nethttps://lists.sourceforge.net/lists/listinfo/factor-talk ---=---Александр -- Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot___ Factor-talk mailing list Factor-talk@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/factor-talk
Re: [Factor-talk] when-empty
Here's a slightly better stack effect documentation: `when-zero ( ..a n quot: ( ..a -- ..b ) -- ..b )` But IIUC the actual stack effect is one of these: `when-zero ( ..a 0 quot: ( ..a -- ..b ) -- ..b )``when-zero ( ..a n#0 quot: ( ..a -- ..b ) -- ..a n )` where # denotes inequality. Am I right? How does the compiler handle such ambiguity in the stack effect?01.03.2017, 19:28, "John Benediktsson" :A stack effect of ( seq quot -- ) doesn't say what the quot is expected to do... You can play around with it and see how it works: IN: scratchpad "abc" [ "empty" ] when-empty . "abc" IN: scratchpad "" [ "empty" ] when-empty . "empty" We should improve the docs here because they are a little hard to follow, since it refers to ``when-empty`` taking the "first quot" of an ``if-empty``, with an empty second quot.On Wed, Mar 1, 2017 at 8:18 AM, Alexander Ilinwrote:Hello! I'm struggling with some of the Factor's words. Mostly with `when-empty`, `if-zero` and similar. It's difficult for me to understand why they don't work sometimes, while their simpler counterparts (while, if) do work exactly as expected. Here's the latest installment of my struggle:IN: scratchpad USE: crypto.xor [ [ empty-xor-key ] when-empty ] infer.( x -- x ) Why is the stack effect as shown here? The documentation for `when-empty` says it takes `( seq quot -- )`. Therefore, I expect `[ [ empty-xor-key ] when-empty ] infer.` to output `( x -- )`. When I modify the quotation passed to `when-empty` to be a no-op, I get a compilation error:IN: scratchpad [ [ 1 drop ] when-empty ] infer.ErrorThe input quotations to “if-empty” don't match their expected effectsInput Expected Got[ 1 drop ] ( ..a -- ..b ) ( -- )[ ] ( ..a seq -- ..b ) ( -- ) That's completely weird to me. Is the stack effect in the documentation incorrect? Should it be `( seq quot -- seq )`? And why does it refuse to compile the second example with `1 drop`?---=--- Александр--Check out the vibrant tech community on one of the world's mostengaging tech sites, SlashDot.org! http://sdm.link/slashdot___Factor-talk mailing listFactor-talk@lists.sourceforge.nethttps://lists.sourceforge.net/lists/listinfo/factor-talk,--Check out the vibrant tech community on one of the world's mostengaging tech sites, SlashDot.org! http://sdm.link/slashdot,___Factor-talk mailing listFactor-talk@lists.sourceforge.nethttps://lists.sourceforge.net/lists/listinfo/factor-talk ---=---Александр -- Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot___ Factor-talk mailing list Factor-talk@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/factor-talk
Re: [Factor-talk] when-empty
A stack effect of ( seq quot -- ) doesn't say what the quot is expected to do... You can play around with it and see how it works: IN: scratchpad "abc" [ "empty" ] when-empty . "abc" IN: scratchpad "" [ "empty" ] when-empty . "empty" We should improve the docs here because they are a little hard to follow, since it refers to ``when-empty`` taking the "first quot" of an ``if-empty``, with an empty second quot. On Wed, Mar 1, 2017 at 8:18 AM, Alexander Ilin wrote: > Hello! > > I'm struggling with some of the Factor's words. Mostly with > `when-empty`, `if-zero` and similar. It's difficult for me to understand > why they don't work sometimes, while their simpler counterparts (while, if) > do work exactly as expected. > > Here's the latest installment of my struggle: > > IN: scratchpad USE: crypto.xor [ [ empty-xor-key ] when-empty ] infer. > ( x -- x ) > > Why is the stack effect as shown here? The documentation for > `when-empty` says it takes `( seq quot -- )`. Therefore, I expect `[ [ > empty-xor-key ] when-empty ] infer.` to output `( x -- )`. > > When I modify the quotation passed to `when-empty` to be a no-op, I get > a compilation error: > IN: scratchpad [ [ 1 drop ] when-empty ] infer. > Error > The input quotations to “if-empty” don't match their expected effects > Input Expected Got > [ 1 drop ] ( ..a -- ..b ) ( -- ) > [ ]( ..a seq -- ..b ) ( -- ) > > That's completely weird to me. Is the stack effect in the documentation > incorrect? Should it be `( seq quot -- seq )`? > And why does it refuse to compile the second example with `1 drop`? > > ---=--- > Александр > > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > ___ > Factor-talk mailing list > Factor-talk@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/factor-talk > -- Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot___ Factor-talk mailing list Factor-talk@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/factor-talk
[Factor-talk] when-empty
Hello! I'm struggling with some of the Factor's words. Mostly with `when-empty`, `if-zero` and similar. It's difficult for me to understand why they don't work sometimes, while their simpler counterparts (while, if) do work exactly as expected. Here's the latest installment of my struggle: IN: scratchpad USE: crypto.xor [ [ empty-xor-key ] when-empty ] infer. ( x -- x ) Why is the stack effect as shown here? The documentation for `when-empty` says it takes `( seq quot -- )`. Therefore, I expect `[ [ empty-xor-key ] when-empty ] infer.` to output `( x -- )`. When I modify the quotation passed to `when-empty` to be a no-op, I get a compilation error: IN: scratchpad [ [ 1 drop ] when-empty ] infer. Error The input quotations to “if-empty” don't match their expected effects Input Expected Got [ 1 drop ] ( ..a -- ..b ) ( -- ) [ ]( ..a seq -- ..b ) ( -- ) That's completely weird to me. Is the stack effect in the documentation incorrect? Should it be `( seq quot -- seq )`? And why does it refuse to compile the second example with `1 drop`? ---=--- Александр -- Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot ___ Factor-talk mailing list Factor-talk@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/factor-talk