Smalltalk › Squeak › Squeak - Dev

Explorer / Inspector bug?

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

18 messages Options

Jaromir Matas

Explorer / Inspector bug?

Hi, I tried to explore and debug this simple code:

p := [[true] whileTrue: [10 second wait]] fork

p explore

... the explorer shows pc=nil and sender=nil but the process is running and
responding correctly to isTerminated test and also showing correct answers
in the bottom mini-workspace. However, when I open another explorer on the
suspendedContext, it shows pc=nil and sender=nil even in the
miniworkspace...

This doesn't seem right but can't figure out where the problem is. And the
explorer has to be run as a separate command...

<http://forum.world.st/file/t372955/Screenshot_2021-03-05_192904.png>

-----
^[^ Jaromir
--
Sent from: http://forum.world.st/Squeak-Dev-f45488.html

^[^ Jaromir

Christoph Thiede

Re: Explorer / Inspector bug?

Hi Jaromir,

from what I know the VM initializes Context objects lazily and only to the required extent. Eliot wrote an interesting "marriage story" about this recently, see: http://forum.world.st/Two-new-curious-Context-primitive-questions-tp5125779p5125782.html

Also, suspendedContext is only set when the process is not running at the moment.

Did you try to fetch the same values from the process itself by using "thisContext"?

Best,

Christoph

Von: Squeak-dev <[hidden email]> im Auftrag von Jaromir Matas <[hidden email]>
Gesendet: Freitag, 5. März 2021 19:37:12
An: [hidden email]
Betreff: [squeak-dev] Explorer / Inspector bug?

Carpe Squeak!

Jaromir Matas

Re: Explorer / Inspector bug?

Hi Christoph,

thanks for the link - it's a brilliant summary of the long detailed version
mentioned inside! Many thanks, Eliot!

Here's my mistake: I setup a `10 second wait` instead of `10 second*s* wait`
and Squeak interprets this typo as 1 second wait so I was watching a history
only - sorry for the noise :)

best,

-----
^[^ Jaromir
--
Sent from: http://forum.world.st/Squeak-Dev-f45488.html

^[^ Jaromir

codefrau

Re: Explorer / Inspector bug?

On Fri, Mar 5, 2021 at 3:07 PM Jaromir Matas <[hidden email]> wrote:

Hi Christoph,

thanks for the link - it's a brilliant summary of the long detailed version
mentioned inside! Many thanks, Eliot!

Hear, hear!

Here's my mistake: I setup a `10 second wait` instead of `10 second*s* wait`
and Squeak interprets this typo as 1 second wait

That sounds like a surprising feature/bug – either it should wait for 10 seconds, or it should throw an error, IMHO, but not simply disregard the receiver. I'd prefer if it just did the expected thing, making #second and #seconds synonyms.

- Vanessa -

so I was watching a history only - sorry for the noise :)

best,

-----
^[^ Jaromir
--
Sent from: http://forum.world.st/Squeak-Dev-f45488.html

Jaromir Matas

Re: Explorer / Inspector bug?

> That sounds like a surprising feature/bug – either it should wait for 10
seconds, or it should throw an error, IMHO, but not simply disregard the
receiver. I'd prefer if it just did the expected thing, making #second and
#seconds synonyms.

+1. It bit me a couple of times already :)

I'm not sure it's grammatically 100% correct but in English there's e.g. a
five-second delay etc. sometimes written as five second delay...

Number >> second

^ self sign seconds

Number >> sign
"Answer 1 if the receiver is greater than 0, -1 if less than 0, else 0."

self > 0 ifTrue: [^1].
self < 0 ifTrue: [^-1].
^0

-----
^[^ Jaromir
--
Sent from: http://forum.world.st/Squeak-Dev-f45488.html

^[^ Jaromir

David T. Lewis

Re: Explorer / Inspector bug?

On Sat, Mar 06, 2021 at 02:34:44AM -0600, Jaromir Matas wrote:
> > That sounds like a surprising feature/bug ??? either it should wait for 10
> seconds, or it should throw an error, IMHO, but not simply disregard the
> receiver. I'd prefer if it just did the expected thing, making #second and
> #seconds synonyms.
>
> +1. It bit me a couple of times already :)
>

+1

My expectation would be that #second, #minute, #hour etc would
be syntactic sugar. And I certainly would not expect "5 second"
to mean one second duration.

I note that we have several methods like this (second, minute, hour...)
and they have Brent Pinkney's (brp) initials since 2004. I don't know
if anyone depends on that existing behavior.

Off topic, but there are other problems with these duration methods.
They are defined in Number but round the duration magnitudes to
whole seconds. And durations for 1 day or 1 week are undefinable
due to daylight savings transitions and occasional leap seconds.

Dave

> I'm not sure it's grammatically 100% correct but in English there's e.g. a
> five-second delay etc. sometimes written as five second delay...
>
>
> Number >> second
>
> ^ self sign seconds
>
> Number >> sign
> "Answer 1 if the receiver is greater than 0, -1 if less than 0, else 0."
>
> self > 0 ifTrue: [^1].
> self < 0 ifTrue: [^-1].
> ^0
>
>
>
>
>
> -----
> ^[^ Jaromir
> --
> Sent from: http://forum.world.st/Squeak-Dev-f45488.html
>

marcel.taeumel

Re: Explorer / Inspector bug?

Hi all!

I think that we should change (and test) all those singular cases. And maybe add lower-case versions for #microSeconds, #milliSeconds, and #nanoSeconds to make it proper English. :-)

Best,

Marcel

Am 06.03.2021 16:05:28 schrieb David T. Lewis <[hidden email]>:
On Sat, Mar 06, 2021 at 02:34:44AM -0600, Jaromir Matas wrote:
> > That sounds like a surprising feature/bug ??? either it should wait for 10
> seconds, or it should throw an error, IMHO, but not simply disregard the
> receiver. I'd prefer if it just did the expected thing, making #second and
> #seconds synonyms.
>
> +1. It bit me a couple of times already :)
>

+1

My expectation would be that #second, #minute, #hour etc would
be syntactic sugar. And I certainly would not expect "5 second"
to mean one second duration.

I note that we have several methods like this (second, minute, hour...)
and they have Brent Pinkney's (brp) initials since 2004. I don't know
if anyone depends on that existing behavior.

Off topic, but there are other problems with these duration methods.
They are defined in Number but round the duration magnitudes to
whole seconds. And durations for 1 day or 1 week are undefinable
due to daylight savings transitions and occasional leap seconds.

Dave

> I'm not sure it's grammatically 100% correct but in English there's e.g. a
> five-second delay etc. sometimes written as five second delay...
>
>
> Number >> second
>
> ^ self sign seconds
>
> Number >> sign
> "Answer 1 if the receiver is greater than 0, -1 if less than 0, else 0."
>
> self > 0 ifTrue: [^1].
> self < 0 ifTrue: [^-1].
> ^0
>
>
>
>
>
> -----
> ^[^ Jaromir
> --
> Sent from: http://forum.world.st/Squeak-Dev-f45488.html
>

Chris Muller-3

Re: Explorer / Inspector bug?

In reply to this post by David T. Lewis

My expectation would be that #second, #minute, #hour etc would
be syntactic sugar. And I certainly would not expect "5 second"
to mean one second duration.

I note that we have several methods like this (second, minute, hour...)
and they have Brent Pinkney's (brp) initials since 2004. I don't know
if anyone depends on that existing behavior.

That's correct. Brent adopted those methods from my MaTimeObjects package (which is still installable from SqueakMap), although in my version it didn't even account for the sign, it was always just ^ 1 second, regardless of the receiver. I agree they should just be synonyms.

Off topic, but there are other problems with these duration methods.
They are defined in Number but round the duration magnitudes to
whole seconds.

Where do you see that? I just tried "1.5 hours" and it worked...

And durations for 1 day or 1 week are undefinable
due to daylight savings transitions and occasional leap seconds.

True, from a literal sense, but by that assessment we would have similar "problems" with ChronologyConstants' DaysInMonth, MicrosecondsInDay, OneDay, SecondsInDay, SecondsInHour, and SecondsInMinute. All are useful, nonetheless.

- Chris

Christoph Thiede

Re: Explorer / Inspector bug?

+1 for preventing "2 second" from answering "1 second", too confusing.

-1 for answering "2 seconds" for "2 second", that would be very ambiguous language and grammatically incorrect. :-)

I think we should either go with an assertion (self assert: self = 1) or deprecate the singular selectors altogether.

Best,

Christoph

Von: Squeak-dev <[hidden email]> im Auftrag von Chris Muller <[hidden email]>
Gesendet: Dienstag, 9. März 2021 05:53:10
An: The general-purpose Squeak developers list
Betreff: Re: [squeak-dev] Explorer / Inspector bug?

My expectation would be that #second, #minute, #hour etc would
be syntactic sugar. And I certainly would not expect "5 second"
to mean one second duration.

I note that we have several methods like this (second, minute, hour...)
and they have Brent Pinkney's (brp) initials since 2004. I don't know
if anyone depends on that existing behavior.

Off topic, but there are other problems with these duration methods.
They are defined in Number but round the duration magnitudes to
whole seconds.

Where do you see that? I just tried "1.5 hours" and it worked...

And durations for 1 day or 1 week are undefinable
due to daylight savings transitions and occasional leap seconds.

- Chris

Carpe Squeak!

Tobias Pape

Re: Explorer / Inspector bug?

>
> I think we should either go with an assertion (self assert: self = 1) or deprecate the singular selectors altogether.
>
> Best,
> Christoph
> Von: Squeak-dev <[hidden email]> im Auftrag von Chris Muller <[hidden email]>
> Gesendet: Dienstag, 9. März 2021 05:53:10
> An: The general-purpose Squeak developers list
> Betreff: Re: [squeak-dev] Explorer / Inspector bug?
>
> My expectation would be that #second, #minute, #hour etc would
> be syntactic sugar. And I certainly would not expect "5 second"
> to mean one second duration.
>
> I note that we have several methods like this (second, minute, hour...)
> and they have Brent Pinkney's (brp) initials since 2004. I don't know
> if anyone depends on that existing behavior.
>
> That's correct. Brent adopted those methods from my MaTimeObjects package (which is still installable from SqueakMap), although in my version it didn't even account for the sign, it was always just ^ 1 second, regardless of the receiver. I agree they should just be synonyms.
>
> Off topic, but there are other problems with these duration methods.
> They are defined in Number but round the duration magnitudes to
> whole seconds.
>
> Where do you see that? I just tried "1.5 hours" and it worked...
>
> And durations for 1 day or 1 week are undefinable
> due to daylight savings transitions and occasional leap seconds.
>
> True, from a literal sense, but by that assessment we would have similar "problems" with ChronologyConstants' DaysInMonth, MicrosecondsInDay, OneDay, SecondsInDay, SecondsInHour, and SecondsInMinute. All are useful, nonetheless.
>
> - Chris

Christoph Thiede

Re: Explorer / Inspector bug?

> "I want a two second delay" is perfectly valid :D

Maybe a native English speaker can correct me but afaik, in the grammar dependency tree of "two-second delay", "two-second" is not an independent noun but a pre-modifying phrase only. Shall we really start to support even more different types of words in an object-oriented language like Smalltalk? Imho, substantives and verbs are already enough. :-)

Otherwise, we might end up with overcomplicated expressions such as "2 seconds later do: [self flash. Wait for: 1.5 hour delay. self flash]". Exaggerating, of course. Though syntactically valid Smalltalk, definitively not "our" flavor of Smalltalk ... Reminds me rather of questionary constructs such as in RSpec (Ruby), for example.

And if this argument is way too complex, here is even a simpler one: According to Grammarly, "two second delay" is not even valid English. It needs to be "two-second delay" and I have never heard of mapping grammatical hyphens to separate messages in Smalltalk languages before. But would like to read some examples if there are any. :-)

Lbnl, I think it causes a lot of confusion and the need for coding styles if we create multiple selectors that do the same thing ... #ifNil:ifNotNil: vs #ifNil:ifNotNilDo:/#ifEmpty:ifNotEmpty: vs #ifEmpty:ifNotEmptyDo: is already hard enough.

Best,

Christoph

Von: Squeak-dev <[hidden email]> im Auftrag von Tobias Pape <[hidden email]>
Gesendet: Dienstag, 9. März 2021 14:26:05
An: The general-purpose Squeak developers list
Betreff: Re: [squeak-dev] Explorer / Inspector bug?

> On 9. Mar 2021, at 14:23, Thiede, Christoph <[hidden email]> wrote:
>
> +1 for preventing "2 second" from answering "1 second", too confusing.

yes

> -1 for answering "2 seconds" for "2 second", that would be very ambiguous language and grammatically incorrect. :-)

no.
"I want a two second delay" is perfectly valid :D

-t

>
> I think we should either go with an assertion (self assert: self = 1) or deprecate the singular selectors altogether.
>
> Best,
> Christoph
> Von: Squeak-dev <[hidden email]> im Auftrag von Chris Muller <[hidden email]>
> Gesendet: Dienstag, 9. März 2021 05:53:10
> An: The general-purpose Squeak developers list
> Betreff: Re: [squeak-dev] Explorer / Inspector bug?
>
> My expectation would be that #second, #minute, #hour etc would
> be syntactic sugar. And I certainly would not expect "5 second"
> to mean one second duration.
>
> I note that we have several methods like this (second, minute, hour...)
> and they have Brent Pinkney's (brp) initials since 2004. I don't know
> if anyone depends on that existing behavior.
>
> That's correct. Brent adopted those methods from my MaTimeObjects package (which is still installable from SqueakMap), although in my version it didn't even account for the sign, it was always just ^ 1 second, regardless of the receiver. I agree they should just be synonyms.
>
> Off topic, but there are other problems with these duration methods.
> They are defined in Number but round the duration magnitudes to
> whole seconds.
>
> Where do you see that? I just tried "1.5 hours" and it worked...
>
> And durations for 1 day or 1 week are undefinable
> due to daylight savings transitions and occasional leap seconds.
>
> True, from a literal sense, but by that assessment we would have similar "problems" with ChronologyConstants' DaysInMonth, MicrosecondsInDay, OneDay, SecondsInDay, SecondsInHour, and SecondsInMinute. All are useful, nonetheless.
>
> - Chris

Carpe Squeak!

David T. Lewis

Re: Explorer / Inspector bug?

On Tue, Mar 09, 2021 at 01:47:49PM +0000, Thiede, Christoph wrote:
> > "I want a two second delay" is perfectly valid :D
>
> Maybe a native English speaker can correct me but afaik, in the grammar
> dependency tree of "two-second delay", "two-second" is not an independet
> noun but a pre-modifying phrase only. Shall we really start to support
> even more different types of words in an object-oriented language like
> Smalltalk? Imho, substantives and verbs are already enough. :-)
>

I am a native Engish speaker with a poor understanding of grammar, so
that makes me a good person to attempt an answer ;-)

The use of singular "second" or plural "seconds" may depend on the
context of the expression, but in any case if I were to hear someone
say "five second" I would understand it to refer to five seconds, not
one second. If I were to try to enforce correct English (is there such
a thing? of course not), then I might try to treat "five second" as
an error. But we do not really need to enforce English syntax rules,
and in this case it is more important for the expression to do the
right thing.

So for me as an English speaker, I overlook the difference between
"second" and "seconds" and I prefer to have the expression do what
I expect in either case.

For this reason, it makes sense to me for #second and #seconds to
be treated as synonyms.

Dave

Nicolas Cellier

Re: Explorer / Inspector bug?

Grammar rules apart, if second has to answer a unit of 1 second, why the hell send such message to 5? Why not Time second for example?

Le jeu. 11 mars 2021 à 02:25, David T. Lewis <[hidden email]> a écrit :

On Tue, Mar 09, 2021 at 01:47:49PM +0000, Thiede, Christoph wrote:
> > "I want a two second delay" is perfectly valid :D
>
> Maybe a native English speaker can correct me but afaik, in the grammar
> dependency tree of "two-second delay", "two-second" is not an independet
> noun but a pre-modifying phrase only. Shall we really start to support
> even more different types of words in an object-oriented language like
> Smalltalk? Imho, substantives and verbs are already enough. :-)
>

I am a native Engish speaker with a poor understanding of grammar, so
that makes me a good person to attempt an answer ;-)

The use of singular "second" or plural "seconds" may depend on the
context of the expression, but in any case if I were to hear someone
say "five second" I would understand it to refer to five seconds, not
one second. If I were to try to enforce correct English (is there such
a thing? of course not), then I might try to treat "five second" as
an error. But we do not really need to enforce English syntax rules,
and in this case it is more important for the expression to do the
right thing.

So for me as an English speaker, I overlook the difference between
"second" and "seconds" and I prefer to have the expression do what
I expect in either case.

For this reason, it makes sense to me for #second and #seconds to
be treated as synonyms.

Dave

Christoph Thiede

Re: Explorer / Inspector bug?

> Grammar rules apart, if second has to answer a unit of 1 second, why the hell send such message to 5? Why not Time second for example?

Just syntactic sugar, similar to #asString instead of "String newFrom: ..." Iirc Kent Beck disrecommended the idea in general, but I like its convenience. :-)

Best,

Christoph

Von: Squeak-dev <[hidden email]> im Auftrag von Nicolas Cellier <[hidden email]>
Gesendet: Donnerstag, 11. März 2021 08:16:51
An: The general-purpose Squeak developers list
Betreff: Re: [squeak-dev] Explorer / Inspector bug?

Grammar rules apart, if second has to answer a unit of 1 second, why the hell send such message to 5? Why not Time second for example?

Le jeu. 11 mars 2021 à 02:25, David T. Lewis <[hidden email]> a écrit :

On Tue, Mar 09, 2021 at 01:47:49PM +0000, Thiede, Christoph wrote:
> > "I want a two second delay" is perfectly valid :D
>
> Maybe a native English speaker can correct me but afaik, in the grammar
> dependency tree of "two-second delay", "two-second" is not an independet
> noun but a pre-modifying phrase only. Shall we really start to support
> even more different types of words in an object-oriented language like
> Smalltalk? Imho, substantives and verbs are already enough. :-)
>

I am a native Engish speaker with a poor understanding of grammar, so
that makes me a good person to attempt an answer ;-)

The use of singular "second" or plural "seconds" may depend on the
context of the expression, but in any case if I were to hear someone
say "five second" I would understand it to refer to five seconds, not
one second. If I were to try to enforce correct English (is there such
a thing? of course not), then I might try to treat "five second" as
an error. But we do not really need to enforce English syntax rules,
and in this case it is more important for the expression to do the
right thing.

So for me as an English speaker, I overlook the difference between
"second" and "seconds" and I prefer to have the expression do what
I expect in either case.

For this reason, it makes sense to me for #second and #seconds to
be treated as synonyms.

Dave

Carpe Squeak!

Nicolas Cellier

Re: Explorer / Inspector bug?

With asString, we ask the object itself to convert as String. It can very well have its own idea how to do it. In case of second, it's a different thing, it acts on behalf of another object, without using the slightest ounce of own identity...

Le jeu. 11 mars 2021 à 11:22, Thiede, Christoph <[hidden email]> a écrit :

> Grammar rules apart, if second has to answer a unit of 1 second, why the hell send such message to 5? Why not Time second for example?

Just syntactic sugar, similar to #asString instead of "String newFrom: ..." Iirc Kent Beck disrecommended the idea in general, but I like its convenience. :-)

Best,

Christoph

Von: Squeak-dev <[hidden email]> im Auftrag von Nicolas Cellier <[hidden email]>
Gesendet: Donnerstag, 11. März 2021 08:16:51
An: The general-purpose Squeak developers list
Betreff: Re: [squeak-dev] Explorer / Inspector bug?

Grammar rules apart, if second has to answer a unit of 1 second, why the hell send such message to 5? Why not Time second for example?

Le jeu. 11 mars 2021 à 02:25, David T. Lewis <[hidden email]> a écrit :

On Tue, Mar 09, 2021 at 01:47:49PM +0000, Thiede, Christoph wrote:
> > "I want a two second delay" is perfectly valid :D
>
> Maybe a native English speaker can correct me but afaik, in the grammar
> dependency tree of "two-second delay", "two-second" is not an independet
> noun but a pre-modifying phrase only. Shall we really start to support
> even more different types of words in an object-oriented language like
> Smalltalk? Imho, substantives and verbs are already enough. :-)
>

I am a native Engish speaker with a poor understanding of grammar, so
that makes me a good person to attempt an answer ;-)

The use of singular "second" or plural "seconds" may depend on the
context of the expression, but in any case if I were to hear someone
say "five second" I would understand it to refer to five seconds, not
one second. If I were to try to enforce correct English (is there such
a thing? of course not), then I might try to treat "five second" as
an error. But we do not really need to enforce English syntax rules,
and in this case it is more important for the expression to do the
right thing.

So for me as an English speaker, I overlook the difference between
"second" and "seconds" and I prefer to have the expression do what
I expect in either case.

For this reason, it makes sense to me for #second and #seconds to
be treated as synonyms.

Dave

Nicolas Cellier

Re: Explorer / Inspector bug?

The only objects whose opinion differ on what second should answer are SequenceableCollection, but that's a case of false polymorphism ;)

Le jeu. 11 mars 2021 à 16:22, Nicolas Cellier <[hidden email]> a écrit :

With asString, we ask the object itself to convert as String. It can very well have its own idea how to do it. In case of second, it's a different thing, it acts on behalf of another object, without using the slightest ounce of own identity...

Le jeu. 11 mars 2021 à 11:22, Thiede, Christoph <[hidden email]> a écrit :

> Grammar rules apart, if second has to answer a unit of 1 second, why the hell send such message to 5? Why not Time second for example?

Just syntactic sugar, similar to #asString instead of "String newFrom: ..." Iirc Kent Beck disrecommended the idea in general, but I like its convenience. :-)

Best,

Christoph

Von: Squeak-dev <[hidden email]> im Auftrag von Nicolas Cellier <[hidden email]>
Gesendet: Donnerstag, 11. März 2021 08:16:51
An: The general-purpose Squeak developers list
Betreff: Re: [squeak-dev] Explorer / Inspector bug?

Grammar rules apart, if second has to answer a unit of 1 second, why the hell send such message to 5? Why not Time second for example?

Le jeu. 11 mars 2021 à 02:25, David T. Lewis <[hidden email]> a écrit :

On Tue, Mar 09, 2021 at 01:47:49PM +0000, Thiede, Christoph wrote:
> > "I want a two second delay" is perfectly valid :D
>
> Maybe a native English speaker can correct me but afaik, in the grammar
> dependency tree of "two-second delay", "two-second" is not an independet
> noun but a pre-modifying phrase only. Shall we really start to support
> even more different types of words in an object-oriented language like
> Smalltalk? Imho, substantives and verbs are already enough. :-)
>

I am a native Engish speaker with a poor understanding of grammar, so
that makes me a good person to attempt an answer ;-)

The use of singular "second" or plural "seconds" may depend on the
context of the expression, but in any case if I were to hear someone
say "five second" I would understand it to refer to five seconds, not
one second. If I were to try to enforce correct English (is there such
a thing? of course not), then I might try to treat "five second" as
an error. But we do not really need to enforce English syntax rules,
and in this case it is more important for the expression to do the
right thing.

So for me as an English speaker, I overlook the difference between
"second" and "seconds" and I prefer to have the expression do what
I expect in either case.

For this reason, it makes sense to me for #second and #seconds to
be treated as synonyms.

Dave

David T. Lewis

Re: Explorer / Inspector bug?

On Thu, Mar 11, 2021 at 04:25:24PM +0100, Nicolas Cellier wrote:
> The only objects whose opinion differ on what second should answer are
> SequenceableCollection, but that's a case of false polymorphism ;)
>

Maybe we should have

1 second ==> 2

;-)

Dave

timrowledge

Re: Explorer / Inspector bug?

> On 2021-03-11, at 9:34 AM, David T. Lewis <[hidden email]> wrote:
>
> On Thu, Mar 11, 2021 at 04:25:24PM +0100, Nicolas Cellier wrote:
>> The only objects whose opinion differ on what second should answer are
>> SequenceableCollection, but that's a case of false polymorphism ;)
>>
>
> Maybe we should have
>
> 1 second ==> 2

I'll second that

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Why use one word when two polysyllabic agglomerates will do?