Smalltalk › Squeak › Squeak - Dev

Two bugs. One of them is a regression.

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

11 messages Options

Mateusz Grotek

Two bugs. One of them is a regression.

I've found two bugs in the current Squeak 4.5 #13712. Please tell me if
they're known. If not I'll create the tickets.

Firstly, there is a regression.
1. Create a class called (for example) ABC.
2. Add an instance variable called (for example) xyz.
3. Open the class comment panel (click on the button "?").
4. The contents of it is:
An ABC is xxxxxxxxx.

Instance Variables
xyz:[TAB][TAB]<Object>

xyz
- xxxxx
5. Remove the [TAB]s
6. Accept it
7. An unrecoverable error shows up (after performing steps above it
keeps appearing when accessing the ABC class):
Error: RemoteString past end of file
The error comes from RemoteString>>text

Secondly there is a funny bug with UserDialogBoxMorph (and possibly
other modal windows).
1. Evaluate: UserDialogBoxMorph inform: 'Hello World!'.
2. The dialog box should be modal, and it is, but one can hack it. Just
drag the "OK" button (but move the mouse outside the button). (mouse
down on the "OK" button, mouse move outside the "OK" button, mouse up).

Other modal windows can be affected too.

marcel.taeumel (old)

Re: Two bugs. One of them is a regression.

The modality in the UserDialogBoxMorph is implemented via holding both mouse and keyboard focus. The halo takes the mouse focus away when growing or dragging a morph--without signaling it to the former focus holder. After that, the modal behavior is lost. However, the dialog is still useful as long as it is in the world because it still triggers a world's sub-cycle.

I guess that it is okay to hack it via halos for debugging reasons. :)

Best,
Marcel

Mateusz Grotek

Re: [SPAM] [squeak-dev] Re: Two bugs. One of them is a regression.

Dnia 12.04.2015 12:33:27, Marcel Taeumel napisał(a):

> The modality in the UserDialogBoxMorph is implemented via holding
> both mouse
> and keyboard focus. The halo takes the mouse focus away when growing
> or
> dragging a morph--without signaling it to the former focus holder.
> After
> that, the modal behavior is lost. However, the dialog is still useful
> as
> long as it is in the world because it still triggers a world's
> sub-cycle.
>
> I guess that it is okay to hack it via halos for debugging reasons. :)
>
> Best,
> Marcel
>

I did not mention the halo, and the bug is there without activating it.
Is it related nevertheless? Of course halos should be allowed to break
the modality. Do halos really have something to do with it?

marcel.taeumel (old)

Re: [SPAM] [squeak-dev] Re: Two bugs. One of them is a regression.

Hmm... how did you then drag the OK button? I assumed via halo. :)

Best,
Marcel

marcel.taeumel (old)

Re: [SPAM] [squeak-dev] Re: Two bugs. One of them is a regression.

Ah, just drag. Literally. :D

Hmmm.....

Best,
Marcel

Chris Muller-3

Re: Two bugs. One of them is a regression.

In reply to this post by marcel.taeumel (old)

On Sun, Apr 12, 2015 at 5:33 AM, Marcel Taeumel
<[hidden email]> wrote:
> The modality in the UserDialogBoxMorph is implemented via holding both mouse
> and keyboard focus. The halo takes the mouse focus away when growing or
> dragging a morph--without signaling it to the former focus holder. After
> that, the modal behavior is lost. However, the dialog is still useful as
> long as it is in the world because it still triggers a world's sub-cycle.

Yeah, it doesn't seem like a very big problem. In fact, now that I
know about it I'll probably find it useful so I _don't_ have to press
the break key..

Its a "feature"! :)

Tobias Pape

Re: Two bugs. One of them is a regression.

In reply to this post by Mateusz Grotek

On 12.04.2015, at 06:16, Mateusz Grotek <[hidden email]> wrote:

> I've found two bugs in the current Squeak 4.5 #13712. Please tell me if they're known. If not I'll create the tickets.
>
> Firstly, there is a regression.
> 1. Create a class called (for example) ABC.
> 2. Add an instance variable called (for example) xyz.
> 3. Open the class comment panel (click on the button "?").
> 4. The contents of it is:
> An ABC is xxxxxxxxx.
>
> Instance Variables
> xyz:[TAB][TAB]<Object>
>
> xyz
> - xxxxx
> 5. Remove the [TAB]s
> 6. Accept it
> 7. An unrecoverable error shows up (after performing steps above it keeps appearing when accessing the ABC class):
> Error: RemoteString past end of file
> The error comes from RemoteString>>text

It may be we know that.
Are you on ubuntu?

Best
-Tobias

Mateusz Grotek

Re: Two bugs. One of them is a regression.

Dnia 12.04.2015 22:11:57, Tobias Pape napisał(a):

>
> On 12.04.2015, at 06:16, Mateusz Grotek <[hidden email]>
> wrote:
>
> > I've found two bugs in the current Squeak 4.5 #13712. Please tell
> me if they're known. If not I'll create the tickets.
> >
> > Firstly, there is a regression.
> > 1. Create a class called (for example) ABC.
> > 2. Add an instance variable called (for example) xyz.
> > 3. Open the class comment panel (click on the button "?").
> > 4. The contents of it is:
> > An ABC is xxxxxxxxx.
> >
> > Instance Variables
> > xyz:[TAB][TAB]<Object>
> >
> > xyz
> > - xxxxx
> > 5. Remove the [TAB]s
> > 6. Accept it
> > 7. An unrecoverable error shows up (after performing steps above it
> keeps appearing when accessing the ABC class):
> > Error: RemoteString past end of file
> > The error comes from RemoteString>>text
>
> It may be we know that.
> Are you on ubuntu?
>
> Best
> -Tobias
>
>
>

No, but on Linux (Gentoo).

Tobias Pape

Re: Two bugs. One of them is a regression.

On 12.04.2015, at 22:33, Mateusz Grotek <[hidden email]> wrote:

> Dnia 12.04.2015 22:11:57, Tobias Pape napisał(a):
>> On 12.04.2015, at 06:16, Mateusz Grotek <[hidden email]> wrote:
>> > I've found two bugs in the current Squeak 4.5 #13712. Please tell me if they're known. If not I'll create the tickets.
>> >
>> > Firstly, there is a regression.
>> > 1. Create a class called (for example) ABC.
>> > 2. Add an instance variable called (for example) xyz.
>> > 3. Open the class comment panel (click on the button "?").
>> > 4. The contents of it is:
>> > An ABC is xxxxxxxxx.
>> >
>> > Instance Variables
>> > xyz:[TAB][TAB]<Object>
>> >
>> > xyz
>> > - xxxxx
>> > 5. Remove the [TAB]s
>> > 6. Accept it
>> > 7. An unrecoverable error shows up (after performing steps above it keeps appearing when accessing the ABC class):
>> > Error: RemoteString past end of file
>> > The error comes from RemoteString>>text
>> It may be we know that.
>> Are you on ubuntu?
>> Best
>> -Tobias
> No, but on Linux (Gentoo).

If you put a

self flush

at the end of WriteStream>>nextChunkPut:
does that help?

best
-Tobias

Mateusz Grotek

Re: Two bugs. One of them is a regression.

Dnia 12.04.2015 23:01:43, Tobias Pape napisał(a):

>
> On 12.04.2015, at 22:33, Mateusz Grotek <[hidden email]>
> wrote:
>
> > Dnia 12.04.2015 22:11:57, Tobias Pape napisał(a):
> >> On 12.04.2015, at 06:16, Mateusz Grotek <[hidden email]>
> wrote:
> >> > I've found two bugs in the current Squeak 4.5 #13712. Please
> tell me if they're known. If not I'll create the tickets.
> >> >
> >> > Firstly, there is a regression.
> >> > 1. Create a class called (for example) ABC.
> >> > 2. Add an instance variable called (for example) xyz.
> >> > 3. Open the class comment panel (click on the button "?").
> >> > 4. The contents of it is:
> >> > An ABC is xxxxxxxxx.
> >> >
> >> > Instance Variables
> >> > xyz:[TAB][TAB]<Object>
> >> >
> >> > xyz
> >> > - xxxxx
> >> > 5. Remove the [TAB]s
> >> > 6. Accept it
> >> > 7. An unrecoverable error shows up (after performing steps above
> it keeps appearing when accessing the ABC class):
> >> > Error: RemoteString past end of file
> >> > The error comes from RemoteString>>text
> >> It may be we know that.
> >> Are you on ubuntu?
> >> Best
> >> -Tobias
> > No, but on Linux (Gentoo).
>
> If you put a
>
> self flush
>
> at the end of WriteStream>>nextChunkPut:
> does that help?
>
> best
> -Tobias
>
>

Yes, that solves the problem.

Best wishes,
Mateusz

Bert Freudenberg

Re: Two bugs. One of them is a regression.

In reply to this post by marcel.taeumel (old)

> On 12.04.2015, at 12:33, Marcel Taeumel <[hidden email]> wrote:
>
> The modality in the UserDialogBoxMorph is implemented via holding both mouse
> and keyboard focus. The halo takes the mouse focus away when growing or
> dragging a morph--without signaling it to the former focus holder. After
> that, the modal behavior is lost. However, the dialog is still useful as
> long as it is in the world because it still triggers a world's sub-cycle.

... but only as long as the current process is running. If you create a new UI process (by error or user interrupt), the dialog will still be in the world but not be able to do anything useful.

- Bert -

smime.p7s (5K) Download Attachment