Pending 4.1 Mantis Bugs

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
11 messages Options
Reply | Threaded
Open this post in threaded view
|

Pending 4.1 Mantis Bugs

Andreas.Raab
Folks -

Even though we were late with getting our tests green, the Mantis status
for the 4.1 bug list looks pretty promising
(http://bugs.squeak.org/view.php?id=7480). I just did a pass to resolve
the issues that could be reasonably resolved and it basically leaves us
with three remaining issues:

[7472] In sq 9563 copying workspace text across images loses line endings
http://bugs.squeak.org/view.php?id=7472

This looks like a Linux / Unix bug. I have absolutely no idea what (if
anything) to do here. Someone please advise.

[7424] the ProportionalSplitterMorph in ArchiveViewer is out of service
http://bugs.squeak.org/view.php?id=7424

This is an interesting bug. Morph>>containingWindow, the root of the
problem used to be a very simply method but got replaced with something
significantly more complex. Does anyone recall what the reason for that
replacement was? It appears to me that replacing all the logic with a
simple "self findOwnerSuchThat:[:m| m isSystemWindow]" ought to be
correct but I'm virtually certain that there must be some reason for the
logic in there and I don't feel very good about replacing some complex
logic like that so shortly before the release. Any insights are welcome.

[7463] In sq 8955 ChangeSorter has a multiple number of ailments
[7342] Change Sorters in 3.10.2 no longer report about class definitions
or class comments.
[7443] In sq 8864 Changesorter shows changeset name twice after loaded
from project.
http://bugs.squeak.org/view.php?id=7463

There is whole bunch of ChangeSorter issues that I am uncertain about.
Some of the comments indicate that portions may have been fixed or
should have been fixed, but I am uncertain what, if any, action to take.

The key question to me here is: Which one of these are show-stoppers?
Which one of these stop you from getting work done, make work results
inaccessible or similar?

In any case, it's nice to see that we're down to five issues to resolve
for 4.1. Thanks everyone for your help!

Cheers,
   - Andreas

Reply | Threaded
Open this post in threaded view
|

Re: Pending 4.1 Mantis Bugs

David T. Lewis
On Wed, Apr 07, 2010 at 08:50:29PM -0700, Andreas Raab wrote:
>
> [7472] In sq 9563 copying workspace text across images loses line endings
> http://bugs.squeak.org/view.php?id=7472
>
> This looks like a Linux / Unix bug. I have absolutely no idea what (if
> anything) to do here. Someone please advise.

Yes, this is something related to the unix only. The issue has existed
for some time, and no fix has been proposed. It should not be considered
a blocker for Squeak 4.1 release.

> In any case, it's nice to see that we're down to five issues to resolve
> for 4.1. Thanks everyone for your help!

Checking again with image now updated to 9920, and running on a Unix VM
version 4.0.2-2172 (compiled as 64-bit), we are now 100% green:

2652 run, 2643 passes, 9 expected failures, 0 failures, 0 errors, 0 unexpected passes

Dave


Reply | Threaded
Open this post in threaded view
|

Re: Pending 4.1 Mantis Bugs

Edgar J. De Cleene-3
In reply to this post by Andreas.Raab
Just I update to Squeak4.1-9924-beta.image

ServerDirectory serverNames
#('Bob SuperSwiki' 'Squeak.org Archive' 'Squeakland Projects' 'Squeakland
SuperSwiki' 'Squeakland-BSS' 'Squeakland-Projects' 'UCSBCreate Archive'
'UIUC Archive').

We must do ServerDirectory  resetServers
Think a empty .mcz with a preamble do the job.

In FileList window, still we not have how to add servers as you see we have
in 4.0
 






CM Capture 1.png (66K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Pending 4.1 Mantis Bugs

laza
In reply to this post by Andreas.Raab
I just went through some issues on mantis (boy there is stuff in there
and it certainly sends you on a roller coaster tour visiting places in
Squeak I never ever have touched).

I added 7453, 7157 just in case, but maybe this is already covered by
the new VMs. Feel free to remove them again from the list.

Alex

2010/4/8 Andreas Raab <[hidden email]>

>
> Folks -
>
> Even though we were late with getting our tests green, the Mantis status for the 4.1 bug list looks pretty promising (http://bugs.squeak.org/view.php?id=7480). I just did a pass to resolve the issues that could be reasonably resolved and it basically leaves us with three remaining issues:
>
> [7472] In sq 9563 copying workspace text across images loses line endings
> http://bugs.squeak.org/view.php?id=7472
>
> This looks like a Linux / Unix bug. I have absolutely no idea what (if anything) to do here. Someone please advise.
>
> [7424] the ProportionalSplitterMorph in ArchiveViewer is out of service
> http://bugs.squeak.org/view.php?id=7424
>
> This is an interesting bug. Morph>>containingWindow, the root of the problem used to be a very simply method but got replaced with something significantly more complex. Does anyone recall what the reason for that replacement was? It appears to me that replacing all the logic with a simple "self findOwnerSuchThat:[:m| m isSystemWindow]" ought to be correct but I'm virtually certain that there must be some reason for the logic in there and I don't feel very good about replacing some complex logic like that so shortly before the release. Any insights are welcome.
>
> [7463] In sq 8955 ChangeSorter has a multiple number of ailments
> [7342] Change Sorters in 3.10.2 no longer report about class definitions or class comments.
> [7443] In sq 8864 Changesorter shows changeset name twice after loaded from project.
> http://bugs.squeak.org/view.php?id=7463
>
> There is whole bunch of ChangeSorter issues that I am uncertain about. Some of the comments indicate that portions may have been fixed or should have been fixed, but I am uncertain what, if any, action to take.
>
> The key question to me here is: Which one of these are show-stoppers? Which one of these stop you from getting work done, make work results inaccessible or similar?
>
> In any case, it's nice to see that we're down to five issues to resolve for 4.1. Thanks everyone for your help!
>
> Cheers,
>  - Andreas
>

Reply | Threaded
Open this post in threaded view
|

Re: Pending 4.1 Mantis Bugs

Edgar J. De Cleene-3



On 4/8/10 12:49 PM, "Alexander Lazarević" <[hidden email]> wrote:

> I just went through some issues on mantis (boy there is stuff in there
> and it certainly sends you on a roller coaster tour visiting places in
> Squeak I never ever have touched).
>
> I added 7453, 7157 just in case, but maybe this is already covered by
> the new VMs. Feel free to remove them again from the list.
>
> Alex


Alex, just receive my old [Squeak 0005628]: Unloading Movies from image from
12-09-2006.

You know 3.10 rip all mp3 related from image ?

No idea if the line commented is still  commented ..
By the way, exist more places in the image referred to classes move out.

Edgar



Reply | Threaded
Open this post in threaded view
|

Re: Pending 4.1 Mantis Bugs

Chris Muller-3
In reply to this post by Andreas.Raab
> Which
> one of these stop you from getting work done, make work results inaccessible
> or similar?

Thanks for asking this, because these are the items that have the
"multiplier" effects on community productivity (which I was trying,
clumsily, to allude to in meeting the other day).

> [7472] In sq 9563 copying workspace text across images loses line endings
> http://bugs.squeak.org/view.php?id=7472
>
> This looks like a Linux / Unix bug. I have absolutely no idea what (if
> anything) to do here. Someone please advise.

I sure wish more folks were experiencing this one, merely for sake of
reproducibility for someone who might know to fix it..   :(

So far, the Mantis report indicates only Jerome, myself and laza have
experienced the problem.

Cut and paste text *between* two running Squeak images loses the
line-endings.  Anyone else out there have that problem?

Here is another question:  Can you copy text in a workspace, exit the
image *without* saving it, re-launch it, and paste the text you
copied?  I can't.  It looks like maybe the clipboard is saved with the
image?

There is one more bit of information about cut and paste;

  - the *first* time I copy text to the clipboard while in Squeak, the
computer beeps at me, but not on subsequent copies to the clipboard
during that image session.  I was wondering whether the beep was in
some error-handler or something.  When I paste the "beeped copied"
text, I get a Beep on the paste too.  Paste paste paste = beep beep
beep.

  - However, if I copy text *again*, it does not beep, and nor do
those pastes beep.  Weird!

Cut and paste seems to work pretty well in my other Linux apps, just
not Squeak..   :(

If it's only me and a few others, this isn't a show-stopper for 4.1.
Hopefully the cause will turn up someday..

Reply | Threaded
Open this post in threaded view
|

Re: Pending 4.1 Mantis Bugs

Casey Ransberger-2
Chris, I can reproduce this issue on my linux box. Also: when I copy, quit w/o saving, relaunch, and paste, nothing happens.

casey@wanderlust:~$ squeak -version
3.10-6 #1 Sun Aug  2 02:02:29 PDT 2009 gcc 4.2.4
Squeak3.10beta of 22 July 2007 [latest update: #7159]
Linux ubuntu 2.6.24-24-generic #1 SMP Fri Jul 24 22:46:06 UTC 2009 i686 GNU/Linux
default plugin location: /usr/local/lib/squeak/3.10-6/*.so


On Fri, Apr 9, 2010 at 2:21 PM, Chris Muller <[hidden email]> wrote:
> Which
> one of these stop you from getting work done, make work results inaccessible
> or similar?

Thanks for asking this, because these are the items that have the
"multiplier" effects on community productivity (which I was trying,
clumsily, to allude to in meeting the other day).

> [7472] In sq 9563 copying workspace text across images loses line endings
> http://bugs.squeak.org/view.php?id=7472
>
> This looks like a Linux / Unix bug. I have absolutely no idea what (if
> anything) to do here. Someone please advise.

I sure wish more folks were experiencing this one, merely for sake of
reproducibility for someone who might know to fix it..   :(

So far, the Mantis report indicates only Jerome, myself and laza have
experienced the problem.

Cut and paste text *between* two running Squeak images loses the
line-endings.  Anyone else out there have that problem?

Here is another question:  Can you copy text in a workspace, exit the
image *without* saving it, re-launch it, and paste the text you
copied?  I can't.  It looks like maybe the clipboard is saved with the
image?

There is one more bit of information about cut and paste;

 - the *first* time I copy text to the clipboard while in Squeak, the
computer beeps at me, but not on subsequent copies to the clipboard
during that image session.  I was wondering whether the beep was in
some error-handler or something.  When I paste the "beeped copied"
text, I get a Beep on the paste too.  Paste paste paste = beep beep
beep.

 - However, if I copy text *again*, it does not beep, and nor do
those pastes beep.  Weird!

Cut and paste seems to work pretty well in my other Linux apps, just
not Squeak..   :(

If it's only me and a few others, this isn't a show-stopper for 4.1.
Hopefully the cause will turn up someday..




Reply | Threaded
Open this post in threaded view
|

Re: Pending 4.1 Mantis Bugs

Ian Piumarta-2
In reply to this post by Chris Muller-3
Chris,

On Apr 9, 2010, at 2:21 PM, Chris Muller wrote:

> [7472] In sq 9563 copying workspace text across images loses line  
> endings

Do you have a valid locale set in your environment?  Try "export  
LC_ALL=C" before starting your images, or maybe "export  
LC_ALL=en_US.UTF-8".

> Here is another question:  Can you copy text in a workspace, exit the
> image *without* saving it, re-launch it, and paste the text you
> copied?  I can't.  It looks like maybe the clipboard is saved with the
> image?

The VM uses the selection mechanism to transfer the clipboard.  When  
you exit the VM the selection becomes disowned and X does not know to  
ask the VM for the selection when you restart it, even if no other app  
has claimed the selection in the meantime.  There are alternative  
mechanisms in Gnome, KDE, etc., but lots of people do not run those  
environments.

FWIW, I cannot reproduce this between Gedit and squeak on Ubuntu 9.10  
-- it works fine for me.  However there might be something in glipper  
that makes the selection mechanism play better with Gnome apps; have  
you tried installing it?

The beeps were for my benefit many years ago.  They are gone now.

Cheers,
Ian


Reply | Threaded
Open this post in threaded view
|

Re: Pending 4.1 Mantis Bugs

Chris Muller-3
> Do you have a valid locale set in your environment?  Try "export LC_ALL=C"

OMG, that looks like it.

After 2 years, I think I can start to use C&P again!  I think I'm
going to cry, thank you!

I just need to make sure my environment set up correctly.  I don't
know anything about localization (I'm American  :-), but I will
certainly do that work.

But I would expect this to come up again and again with other
Linux-dummies like me.  Whatever the different path of code inside the
VM is taken on account of this Environment variable not being set; is
there any way to have C&P follow the default expected behavior even if
LC_ALL is not set?




On Fri, Apr 9, 2010 at 5:21 PM, Ian Piumarta <[hidden email]> wrote:

> Chris,
>
> On Apr 9, 2010, at 2:21 PM, Chris Muller wrote:
>
>> [7472] In sq 9563 copying workspace text across images loses line endings
>
> Do you have a valid locale set in your environment?  Try "export LC_ALL=C"
> before starting your images, or maybe "export LC_ALL=en_US.UTF-8".
>
>> Here is another question:  Can you copy text in a workspace, exit the
>> image *without* saving it, re-launch it, and paste the text you
>> copied?  I can't.  It looks like maybe the clipboard is saved with the
>> image?
>
> The VM uses the selection mechanism to transfer the clipboard.  When you
> exit the VM the selection becomes disowned and X does not know to ask the VM
> for the selection when you restart it, even if no other app has claimed the
> selection in the meantime.  There are alternative mechanisms in Gnome, KDE,
> etc., but lots of people do not run those environments.
>
> FWIW, I cannot reproduce this between Gedit and squeak on Ubuntu 9.10 -- it
> works fine for me.  However there might be something in glipper that makes
> the selection mechanism play better with Gnome apps; have you tried
> installing it?
>
> The beeps were for my benefit many years ago.  They are gone now.
>
> Cheers,
> Ian
>
>

Reply | Threaded
Open this post in threaded view
|

Re: Pending 4.1 Mantis Bugs

Bert Freudenberg
On 10.04.2010, at 01:33, Chris Muller wrote:

>
>> Do you have a valid locale set in your environment?  Try "export LC_ALL=C"
>
> OMG, that looks like it.
>
> After 2 years, I think I can start to use C&P again!  I think I'm
> going to cry, thank you!
>
> I just need to make sure my environment set up correctly.  I don't
> know anything about localization (I'm American  :-), but I will
> certainly do that work.
>
> But I would expect this to come up again and again with other
> Linux-dummies like me.  Whatever the different path of code inside the
> VM is taken on account of this Environment variable not being set; is
> there any way to have C&P follow the default expected behavior even if
> LC_ALL is not set?
>
>
> On Fri, Apr 9, 2010 at 5:21 PM, Ian Piumarta <[hidden email]> wrote:
>> Do you have a valid locale set in your environment?  Try "export LC_ALL=C"
>> before starting your images, or maybe "export LC_ALL=en_US.UTF-8".


In the Etoys start script we have

[ -z "$LC_ALL" ] && export LC_ALL="$LANG"

so possibly that's why it has been working ...

- Bert -


Reply | Threaded
Open this post in threaded view
|

Re: Pending 4.1 Mantis Bugs

Ian Piumarta-2
On Apr 10, 2010, at 9:24 AM, Bert Freudenberg wrote:

> In the Etoys start script we have
>
> [ -z "$LC_ALL" ] && export LC_ALL="$LANG"
>
> so possibly that's why it has been working ...

It's stranger than that.  If LANG is not set then EOL works right even  
when LC_* are unset.  The situation that causes problems is LANG set  
and LC_* not set.  setenv()ing LC_* based on the result of  
setlocal("") (which usually returns LANG when LC_* are not set) does  
not fix the broken behaviour.  Something in iconv is set early in  
process startup, ignores calls to setlocale(), and refuses to be fixed  
later during execution.  The script solution seems the simplest and  
neatest way to address this.

Cheers,
Ian