Mac + Cog - mouse clicks

> Hello,
>
> i were able to successfully build Cog VM on Mac using freshly
> generated cmake config.
> I can build it using ordinal makefiles as well as using xcode project
> (cmake can generate both).
>
> And everything works ok, except a little nuisance: images are deaf to
> mouse clicks!
> Note, i am talking exactly about mouse clicks, because everything else
> (keys and mouse move events are handled adequately).
>
> Now, does anyone met such issue before? Any clues?
>
> I can only guess, that i'm either miss some obscure #define /#ifdef ,
> or having a wrong includes search path or link order.
>
>
> --
> Best regards,
> Igor Stasenko AKA sig.
>

> Okay, apparently its because i didn't bundled binary into '.app' thingy
> and so, it doens't have an associated .pinfo file where some settings
> related to mouse buttons (i think) are provided. There are some
> 'SqueakMouseXXXButtonYY' values, which seems like responsible for mouse clicks.
> (Hmm, why binary doesn't carry a default ones, when values in .pinfo
> is missing?)
>
> I found the way how to tell cmake to create a bundle,
> and now even worse. It assumes that it budled , and tries to read
> missing keys and use them,
> causing NULL-pointer exception..
> Because i didn't provided correct .plist yet, but using default one,
> which cmake generates.
>

>
> On 24 December 2010 01:40, Igor Stasenko <[hidden email]> wrote:
>> Okay, apparently its because i didn't bundled binary into '.app' thingy
>> and so, it doens't have an associated .pinfo file where some settings
>> related to mouse buttons (i think) are provided. There are some
>> 'SqueakMouseXXXButtonYY' values, which seems like responsible for mouse clicks.
>> (Hmm, why binary doesn't carry a default ones, when values in .pinfo
>> is missing?)
>>
>> I found the way how to tell cmake to create a bundle,
>> and now even worse. It assumes that it budled , and tries to read
>> missing keys and use them,
>> causing NULL-pointer exception..
>> Because i didn't provided correct .plist yet, but using default one,
>> which cmake generates.
>>
> .. and after borrowing 'correct' .pinfo from other cog app, and image
> starts and everything working normally.
>
> :)
>
> --
> Best regards,
> Igor Stasenko AKA sig.

>
> yeah, but we should have some defaults... application shouldn't crash just because some guy removed parameters from pInfo... I'm going to check this in Cocoa Cog :P
>
> Cheers,
> Esteban
>
> El 23/12/2010, a las 9:49p.m., Igor Stasenko escribió:
>
>>
>> On 24 December 2010 01:40, Igor Stasenko <[hidden email]> wrote:
>>> Okay, apparently its because i didn't bundled binary into '.app' thingy
>>> and so, it doens't have an associated .pinfo file where some settings
>>> related to mouse buttons (i think) are provided. There are some
>>> 'SqueakMouseXXXButtonYY' values, which seems like responsible for mouse clicks.
>>> (Hmm, why binary doesn't carry a default ones, when values in .pinfo
>>> is missing?)
>>>
>>> I found the way how to tell cmake to create a bundle,
>>> and now even worse. It assumes that it budled , and tries to read
>>> missing keys and use them,
>>> causing NULL-pointer exception..
>>> Because i didn't provided correct .plist yet, but using default one,
>>> which cmake generates.
>>>
>> .. and after borrowing 'correct' .pinfo from other cog app, and image
>> starts and everything working normally.
>>
>> :)
>>
>> --
>> Best regards,
>> Igor Stasenko AKA sig.
>

> On 2010-12-23, at 5:12 PM, Esteban Lorenzano wrote:
>
>>
>> yeah, but we should have some defaults... application shouldn't crash just because some guy removed parameters from pInfo... I'm going to check this in Cocoa Cog :P
>>
>> Cheers,
>> Esteban
>>
>> El 23/12/2010, a las 9:49p.m., Igor Stasenko escribió:
>>
>>>
>>> On 24 December 2010 01:40, Igor Stasenko <[hidden email]> wrote:
>>>> Okay, apparently its because i didn't bundled binary into '.app' thingy
>>>> and so, it doens't have an associated .pinfo file where some settings
>>>> related to mouse buttons (i think) are provided. There are some
>>>> 'SqueakMouseXXXButtonYY' values, which seems like responsible for mouse clicks.
>>>> (Hmm, why binary doesn't carry a default ones, when values in .pinfo
>>>> is missing?)
>>>>
>>>> I found the way how to tell cmake to create a bundle,
>>>> and now even worse. It assumes that it budled , and tries to read
>>>> missing keys and use them,
>>>> causing NULL-pointer exception..
>>>> Because i didn't provided correct .plist yet, but using default one,
>>>> which cmake generates.
>>>>
>>> .. and after borrowing 'correct' .pinfo from other cog app, and image
>>> starts and everything working normally.
>>>
>>> :)
>>>
>>> --
>>> Best regards,
>>> Igor Stasenko AKA sig.
>>
>
> --
> ===========================================================================
> John M. McIntosh <[hidden email]>   Twitter:  squeaker68882
> Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
> ===========================================================================
>
>
>
>
>

> On 24 December 2010 02:16, John M McIntosh
> <[hidden email]> wrote:
>>
>> Historically I had not put the defaults in the binary because people sometimes would grab the binary and toss all the other parts of the bundle.  Later no one understood how to change a default because the plist hint was missing.
>>
> Yeah.. well.. to prevent tossing, i would simply bummer with message
> like 'hey! where is my bundle?'. :)
>
> It is easy to do by applying following pattern everywhere:
>
> replace all calls like
> value := getKeyValue( KeyName)
> with:
> value := getKeyValue( KeyName , defaultValue)
>
> so, there is no chance to get NULL & crash, because returned values
> are always valid.
>
> Esteban, it all yours..
> I don't have time doing it right now, because i got like 6 other
> configs to write :)
>

> On 2010-12-23, at 5:46 PM, Igor Stasenko wrote:
>
>> On 24 December 2010 02:16, John M McIntosh
>> <[hidden email]> wrote:
>>>
>>> Historically I had not put the defaults in the binary because people sometimes would grab the binary and toss all the other parts of the bundle.  Later no one understood how to change a default because the plist hint was missing.
>>>
>> Yeah.. well.. to prevent tossing, i would simply bummer with message
>> like 'hey! where is my bundle?'. :)
>>
>> It is easy to do by applying following pattern everywhere:
>>
>> replace all calls like
>> value := getKeyValue( KeyName)
>> with:
>> value := getKeyValue( KeyName , defaultValue)
>>
>> so, there is no chance to get NULL & crash, because returned values
>> are always valid.
>>
>> Esteban, it all yours..
>> I don't have time doing it right now, because i got like 6 other
>> configs to write :)
>>
>
> --
> ===========================================================================
> John M. McIntosh <[hidden email]>   Twitter:  squeaker68882
> Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
> ===========================================================================
>
>
>
>
>

> On 24 December 2010 02:51, John M McIntosh
> <[hidden email]> wrote:
>> Oh and btw you'll need to look into the signing logic, I'm not quite sure where that gem of copy protection sits but don't be surprised if
>> Apple pursues it a bit more aggressively...
>>
>
> well. i cut it out of my Cog on mac os.. because one don't needs a license.
> On iOS does.. but Cog doesn't runs there :)
> But correct me if i wrong.
>
>> mmm other thoughts.
>> The plugins *could* be just the raw binaries ala unix builds. The only reason they were bundles is because in early 2000 era we ran os-9 and os-x flavors.
>> That could be all dropped, and maybe just forgo creating plugin bundles.
>>
>>
> Thanks for details, i added an issue ticket to not forget this.
> (http://code.google.com/p/pharo/issues/detail?id=3480)

>> On 2010-12-23, at 5:46 PM, Igor Stasenko wrote:
>>
>>> On 24 December 2010 02:16, John M McIntosh
>>> <[hidden email]>  wrote:
>>>>
>>>> Historically I had not put the defaults in the binary because people sometimes would grab the binary and toss all the other parts of the bundle.  Later no one understood how to change a default because the plist hint was missing.
>>>>
>>> Yeah.. well.. to prevent tossing, i would simply bummer with message
>>> like 'hey! where is my bundle?'. :)
>>>
>>> It is easy to do by applying following pattern everywhere:
>>>
>>> replace all calls like
>>> value := getKeyValue( KeyName)
>>> with:
>>> value := getKeyValue( KeyName , defaultValue)
>>>
>>> so, there is no chance to get NULL&  crash, because returned values
>>> are always valid.
>>>
>>> Esteban, it all yours..
>>> I don't have time doing it right now, because i got like 6 other
>>> configs to write :)
>>>
>>
>> --
>> ===========================================================================
>> John M. McIntosh<[hidden email]>     Twitter:  squeaker68882
>> Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
>> ===========================================================================
>>
>>
>>
>>
>>
>
>
>

> On 2010-12-23, at 6:30 PM, Igor Stasenko wrote:
>
>> On 24 December 2010 02:51, John M McIntosh
>> <[hidden email]> wrote:
>>> Oh and btw you'll need to look into the signing logic, I'm not quite sure where that gem of copy protection sits but don't be surprised if
>>> Apple pursues it a bit more aggressively...
>>>
>>
>> well. i cut it out of my Cog on mac os.. because one don't needs a license.
>> On iOS does.. but Cog doesn't runs there :)
>> But correct me if i wrong.
>>
>>> mmm other thoughts.
>>> The plugins *could* be just the raw binaries ala unix builds. The only reason they were bundles is because in early 2000 era we ran os-9 and os-x flavors.
>>> That could be all dropped, and maybe just forgo creating plugin bundles.
>>>
>>>
>> Thanks for details, i added an issue ticket to not forget this.
>> (http://code.google.com/p/pharo/issues/detail?id=3480)
>
> --
> ===========================================================================
> John M. McIntosh <[hidden email]>   Twitter:  squeaker68882
> Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
> ===========================================================================
>
>
>
>
>

>
> On 24 December 2010 03:41, John M McIntosh
> <[hidden email]> wrote:
>> Apple is moving towards code signing for all binaries. The impact of that on independent developers is unknown. However in order to distribute squeak or a cog vm via the os-x app store everything has to be signed. This includes the Vm and all the plugin binaries.  One can choose to distribute outside the os-x app store, but I think? the community would like to see squeak and cog in the os-x store?
>>
>
> Well, i am new on the Mac field.
> I hope Esteban having a developer license to sign our official
> binaries in future, because i'm not in the mood of becoming a
> 'licensed' developer.
> I am just a hacker who passing by :)
>
>> On 2010-12-23, at 6:30 PM, Igor Stasenko wrote:
>>
>>> On 24 December 2010 02:51, John M McIntosh
>>> <[hidden email]> wrote:
>>>> Oh and btw you'll need to look into the signing logic, I'm not quite sure where that gem of copy protection sits but don't be surprised if
>>>> Apple pursues it a bit more aggressively...
>>>>
>>>
>>> well. i cut it out of my Cog on mac os.. because one don't needs a license.
>>> On iOS does.. but Cog doesn't runs there :)
>>> But correct me if i wrong.
>>>
>>>> mmm other thoughts.
>>>> The plugins *could* be just the raw binaries ala unix builds. The only reason they were bundles is because in early 2000 era we ran os-9 and os-x flavors.
>>>> That could be all dropped, and maybe just forgo creating plugin bundles.
>>>>
>>>>
>>> Thanks for details, i added an issue ticket to not forget this.
>>> (http://code.google.com/p/pharo/issues/detail?id=3480)
>>
>> --
>> ===========================================================================
>> John M. McIntosh <[hidden email]>   Twitter:  squeaker68882
>> Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
>> ===========================================================================
>>
>>
>>
>>
>>
>
>
>
> --
> Best regards,
> Igor Stasenko AKA sig.

> On 24 December 2010 03:41, John M McIntosh
> <[hidden email]> wrote:
>> Apple is moving towards code signing for all binaries. The impact of that on independent developers is unknown. However in order to distribute squeak or a cog vm via the os-x app store everything has to be signed. This includes the Vm and all the plugin binaries.  One can choose to distribute outside the os-x app store, but I think? the community would like to see squeak and cog in the os-x store?
>>
>
> Well, i am new on the Mac field.
> I hope Esteban having a developer license to sign our official
> binaries in future, because i'm not in the mood of becoming a
> 'licensed' developer.
> I am just a hacker who passing by :)
>
>> On 2010-12-23, at 6:30 PM, Igor Stasenko wrote:
>>
>>> On 24 December 2010 02:51, John M McIntosh
>>> <[hidden email]> wrote:
>>>> Oh and btw you'll need to look into the signing logic, I'm not quite sure where that gem of copy protection sits but don't be surprised if
>>>> Apple pursues it a bit more aggressively...
>>>>
>>>
>>> well. i cut it out of my Cog on mac os.. because one don't needs a license.
>>> On iOS does.. but Cog doesn't runs there :)
>>> But correct me if i wrong.
>>>
>>>> mmm other thoughts.
>>>> The plugins *could* be just the raw binaries ala unix builds. The only reason they were bundles is because in early 2000 era we ran os-9 and os-x flavors.
>>>> That could be all dropped, and maybe just forgo creating plugin bundles.
>>>>
>>>>
>>> Thanks for details, i added an issue ticket to not forget this.
>>> (http://code.google.com/p/pharo/issues/detail?id=3480)
>>
>> --
>> ===========================================================================
>> John M. McIntosh <[hidden email]>   Twitter:  squeaker68882
>> Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
>> ===========================================================================
>>
>>
>>
>>
>>
>
>
>
> --
> Best regards,
> Igor Stasenko AKA sig.

>
> On 12/23/2010 6:30 PM, Igor Stasenko wrote:
>>>
>>> mmm other thoughts.
>>> The plugins *could* be just the raw binaries ala unix builds. The only
>>> reason they were bundles is because in early 2000 era we ran os-9 and os-x
>>> flavors.
>>> That could be all dropped, and maybe just forgo creating plugin bundles.
>>>
>>>
>> Thanks for details, i added an issue ticket to not forget this.
>> (http://code.google.com/p/pharo/issues/detail?id=3480)
>
> I think it would be better to file general VM issues on bugs.squeak.org.
>

> Thanks,
>  - Andreas
>
>>> On 2010-12-23, at 5:46 PM, Igor Stasenko wrote:
>>>
>>>> On 24 December 2010 02:16, John M McIntosh
>>>> <[hidden email]>  wrote:
>>>>>
>>>>> Historically I had not put the defaults in the binary because people
>>>>> sometimes would grab the binary and toss all the other parts of the bundle.
>>>>>  Later no one understood how to change a default because the plist hint was
>>>>> missing.
>>>>>
>>>> Yeah.. well.. to prevent tossing, i would simply bummer with message
>>>> like 'hey! where is my bundle?'. :)
>>>>
>>>> It is easy to do by applying following pattern everywhere:
>>>>
>>>> replace all calls like
>>>> value := getKeyValue( KeyName)
>>>> with:
>>>> value := getKeyValue( KeyName , defaultValue)
>>>>
>>>> so, there is no chance to get NULL&  crash, because returned values
>>>> are always valid.
>>>>
>>>> Esteban, it all yours..
>>>> I don't have time doing it right now, because i got like 6 other
>>>> configs to write :)
>>>>
>>>
>>> --
>>>
>>> ===========================================================================
>>> John M. McIntosh<[hidden email]>     Twitter:
>>>  squeaker68882
>>> Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
>>>
>>> ===========================================================================
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>

> On 24 December 2010 03:55, John M McIntosh
> <[hidden email]> wrote:
>> Well the community needs to start thinking about the cost associated with supporting the VM building.
>> Not only will in the future you need a $99 license to sign os-x VMs, the intel compiler just produces better code.
>> This might change over time with LLVM, but right now the LLVM compile won't create a VM. This perhaps is a step UP from
>> the past where LLVM would produce a VM that crashes, or is it?
>>
>> But the entry cost of reporting a bug is the $99 license.  Oh well MAYBE one can do it for free, but good luck with that...
>>
> yeah and this feels a bit alienating..
> "For privilege to work with us, and develop under our superb system,
> you must pay the ransom".. Something wrong with these guys..
> but enough of that.
>
>> On 2010-12-23, at 6:51 PM, Igor Stasenko wrote:
>>
>
> --
> Best regards,
> Igor Stasenko AKA sig.

> On 24 December 2010 03:46, Andreas Raab<[hidden email]>  wrote:
>> On 12/23/2010 6:30 PM, Igor Stasenko wrote:
>>>>
>>>> mmm other thoughts.
>>>> The plugins *could* be just the raw binaries ala unix builds. The only
>>>> reason they were bundles is because in early 2000 era we ran os-9 and os-x
>>>> flavors.
>>>> That could be all dropped, and maybe just forgo creating plugin bundles.
>>>>
>>>>
>>> Thanks for details, i added an issue ticket to not forget this.
>>> (http://code.google.com/p/pharo/issues/detail?id=3480)
>>
>> I think it would be better to file general VM issues on bugs.squeak.org.
>>
> done (http://bugs.squeak.org/view.php?id=7590).
> i put it simply put it as a reminder for my self, because considered
> John's comments worth it.

>>>> On 2010-12-23, at 5:46 PM, Igor Stasenko wrote:
>>>>
>>>>> On 24 December 2010 02:16, John M McIntosh
>>>>> <[hidden email]>    wrote:
>>>>>>
>>>>>> Historically I had not put the defaults in the binary because people
>>>>>> sometimes would grab the binary and toss all the other parts of the bundle.
>>>>>>   Later no one understood how to change a default because the plist hint was
>>>>>> missing.
>>>>>>
>>>>> Yeah.. well.. to prevent tossing, i would simply bummer with message
>>>>> like 'hey! where is my bundle?'. :)
>>>>>
>>>>> It is easy to do by applying following pattern everywhere:
>>>>>
>>>>> replace all calls like
>>>>> value := getKeyValue( KeyName)
>>>>> with:
>>>>> value := getKeyValue( KeyName , defaultValue)
>>>>>
>>>>> so, there is no chance to get NULL&    crash, because returned values
>>>>> are always valid.
>>>>>
>>>>> Esteban, it all yours..
>>>>> I don't have time doing it right now, because i got like 6 other
>>>>> configs to write :)
>>>>>
>>>>
>>>> --
>>>>
>>>> ===========================================================================
>>>> John M. McIntosh<[hidden email]>       Twitter:
>>>>   squeaker68882
>>>> Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
>>>>
>>>> ===========================================================================
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>
>
>
>

>
> $$$$$$$$$$$$$$$$$
>
> So it has come to pass. Steve Jobs, the polo-necked Apple chief, described this week by US president Barack Obama as the “epitome of the American dream”, is the Financial Times’ Person of the Year.
>
> http://www.ft.com/cms/s/0/a420b564-0eef-11e0-9ec3-00144feabdc0.html
>
> {I would not be surprised if you can view that, it's all about $$$}
>
> And you too can pay the fee...
>
> Seriously tho I think the industry is moving towards signing binaries.

> A little known fact is that I PGP signed binaries for years now, look for the Squeak VM Opt.sig which is a PGP signature for the binary, people are welcome to confirm the binary they have is actually signed by me...  I've viewed that as a security measure.  Never signed plugins, just too much hassle.
>
>
> On 2010-12-23, at 7:01 PM, Igor Stasenko wrote:
>
>> On 24 December 2010 03:55, John M McIntosh
>> <[hidden email]>  wrote:
>>> Well the community needs to start thinking about the cost associated with supporting the VM building.
>>> Not only will in the future you need a $99 license to sign os-x VMs, the intel compiler just produces better code.
>>> This might change over time with LLVM, but right now the LLVM compile won't create a VM. This perhaps is a step UP from
>>> the past where LLVM would produce a VM that crashes, or is it?
>>>
>>> But the entry cost of reporting a bug is the $99 license.  Oh well MAYBE one can do it for free, but good luck with that...
>>>
>> yeah and this feels a bit alienating..
>> "For privilege to work with us, and develop under our superb system,
>> you must pay the ransom".. Something wrong with these guys..
>> but enough of that.
>>
>>> On 2010-12-23, at 6:51 PM, Igor Stasenko wrote:
>>>
>>
>> --
>> Best regards,
>> Igor Stasenko AKA sig.
>
> --
> ===========================================================================
> John M. McIntosh<[hidden email]>    Twitter:  squeaker68882
> Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
> ===========================================================================
>
>
>
>
>