Re: [Pharo-project] Fuel for Squeak stupid questions
On Tue, Oct 16, 2012 at 2:06 PM, Max Leske maxle...@gmail.com wrote: Edgar, There is indeed a bug in Fuel, namely Fuel ignores the local time zone. The reason why I didn't detect that in my image is that Squeak seems to have some problems of its own in determining the correct time zone… :) Mm is this Squeak only or also in Pharo? I'll write a fix for that. Cheers, Max On 16.10.2012, at 12:55, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/16/12 3:52 AM, Max Leske maxle...@gmail.com wrote: I checked against build 11925 (found here: http://ftp.squeak.org/trunk/). Did you get your image from somewhere else? I can take a look at the image if you like. You'd just have to put it somewhere for me to download (together with the changes of course). Max Changes was grouped into 'SL' ChangeSet These days try to learn from Fuel and from Phobos to get into this image, currently I use modify HV2 as light web framework The .image is saved as NO Cog format and runs with both Elliot and Esteban or with old John, in my case I have old hardware (that's why I saved and updated with Squeak 4.2.5beta1U ) I send direct to you as this is very preliminar for going public Very thanks Edgar -- Mariano http://marianopeck.wordpress.com
Re: [Pharo-project] Fuel for Squeak stupid questions
The Fuel bug also exists in Pharo (Fuel simply doesn't write out the local offset information) but the timezones seem to be fine. On 17.10.2012, at 12:32, Mariano Martinez Peck marianop...@gmail.com wrote: On Tue, Oct 16, 2012 at 2:06 PM, Max Leske maxle...@gmail.com wrote: Edgar, There is indeed a bug in Fuel, namely Fuel ignores the local time zone. The reason why I didn't detect that in my image is that Squeak seems to have some problems of its own in determining the correct time zone… :) Mm is this Squeak only or also in Pharo? I'll write a fix for that. Cheers, Max On 16.10.2012, at 12:55, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/16/12 3:52 AM, Max Leske maxle...@gmail.com wrote: I checked against build 11925 (found here: http://ftp.squeak.org/trunk/). Did you get your image from somewhere else? I can take a look at the image if you like. You'd just have to put it somewhere for me to download (together with the changes of course). Max Changes was grouped into 'SL' ChangeSet These days try to learn from Fuel and from Phobos to get into this image, currently I use modify HV2 as light web framework The .image is saved as NO Cog format and runs with both Elliot and Esteban or with old John, in my case I have old hardware (that's why I saved and updated with Squeak 4.2.5beta1U ) I send direct to you as this is very preliminar for going public Very thanks Edgar -- Mariano http://marianopeck.wordpress.com
Re: [Pharo-project] Fuel for Squeak stupid questions
I checked against build 11925 (found here: http://ftp.squeak.org/trunk/). Did you get your image from somewhere else? I can take a look at the image if you like. You'd just have to put it somewhere for me to download (together with the changes of course). Max On 15.10.2012, at 22:19, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/15/12 5:10 PM, Max Leske maxle...@gmail.com wrote: Hey Edgar I checked the Fuel tests in a 4.4 image and there are no differences to the 4.3 image. If you still have that problem with the failing Date test then you probably have an issue with the code you loaded in addition. One package that would break that test is PostgresV2 because that packages overrides #year:month:day: in Date and DateAndTime. Cheers, Max I re check, do not remember any I made change Date. Thanks Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
Edgar, There is indeed a bug in Fuel, namely Fuel ignores the local time zone. The reason why I didn't detect that in my image is that Squeak seems to have some problems of its own in determining the correct time zone… :) I'll write a fix for that. Cheers, Max On 16.10.2012, at 12:55, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/16/12 3:52 AM, Max Leske maxle...@gmail.com wrote: I checked against build 11925 (found here: http://ftp.squeak.org/trunk/). Did you get your image from somewhere else? I can take a look at the image if you like. You'd just have to put it somewhere for me to download (together with the changes of course). Max Changes was grouped into 'SL' ChangeSet These days try to learn from Fuel and from Phobos to get into this image, currently I use modify HV2 as light web framework The .image is saved as NO Cog format and runs with both Elliot and Esteban or with old John, in my case I have old hardware (that's why I saved and updated with Squeak 4.2.5beta1U ) I send direct to you as this is very preliminar for going public Very thanks Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
On 10/16/12 9:06 AM, Max Leske maxle...@gmail.com wrote: Edgar, There is indeed a bug in Fuel, namely Fuel ignores the local time zone. The reason why I didn't detect that in my image is that Squeak seems to have some problems of its own in determining the correct time zone :) I'll write a fix for that. Cheers, Max I remember some discuss in Squeak list about this Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
Hey Edgar I checked the Fuel tests in a 4.4 image and there are no differences to the 4.3 image. If you still have that problem with the failing Date test then you probably have an issue with the code you loaded in addition. One package that would break that test is PostgresV2 because that packages overrides #year:month:day: in Date and DateAndTime. Cheers, Max On 12.10.2012, at 08:33, Max Leske maxle...@gmail.com wrote: Thanks Edgar. Much appreciated. On 11.10.2012, at 21:14, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/11/12 12:53 PM, Max Leske maxle...@gmail.com wrote: Ah! Could you try the Date fix I proposed in a 4.3 image? BTW I'm working on fixing the other failing tests (for 4.3) aswell. Cheers, Max You was right, some changed recently in Squeak as 4.3 11860 (fresh unzipped) do not fail the two test of Date, the rest is same as 4.4 You only need fileIn TClassAndTraitDescription from Pharo when Monticello complains Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
On 10/15/12 5:10 PM, Max Leske maxle...@gmail.com wrote: Hey Edgar I checked the Fuel tests in a 4.4 image and there are no differences to the 4.3 image. If you still have that problem with the failing Date test then you probably have an issue with the code you loaded in addition. One package that would break that test is PostgresV2 because that packages overrides #year:month:day: in Date and DateAndTime. Cheers, Max I re check, do not remember any I made change Date. Thanks Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
Thanks Edgar. Much appreciated. On 11.10.2012, at 21:14, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/11/12 12:53 PM, Max Leske maxle...@gmail.com wrote: Ah! Could you try the Date fix I proposed in a 4.3 image? BTW I'm working on fixing the other failing tests (for 4.3) aswell. Cheers, Max You was right, some changed recently in Squeak as 4.3 11860 (fresh unzipped) do not fail the two test of Date, the rest is same as 4.4 You only need fileIn TClassAndTraitDescription from Pharo when Monticello complains Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
On 10/10/12 11:04 AM, Max Leske maxle...@gmail.com wrote: Edgar, which version of Squeak are you using? Max Current Squeak 4.4 updated with some code I add. Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
Ah! Could you try the Date fix I proposed in a 4.3 image? BTW I'm working on fixing the other failing tests (for 4.3) aswell. Cheers, Max On 11.10.2012, at 11:48, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/10/12 11:04 AM, Max Leske maxle...@gmail.com wrote: Edgar, which version of Squeak are you using? Max Current Squeak 4.4 updated with some code I add. Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
On Thu, Oct 11, 2012 at 5:53 PM, Max Leske maxle...@gmail.com wrote: Ah! Could you try the Date fix I proposed in a 4.3 image? BTW I'm working on fixing the other failing tests (for 4.3) aswell. Excellent. Once you finish, let us know. We can do a compatibility package if you want. It works like this for Pharo: last Fuel (trunk) is supposed to work with last (trunk) of Pharo, right now 2.0. Then, we have a package with overrides or some other dirty hacks to work with each previous version of Pharo. And then Metacello takes care of automatically load the necessary compatibility packages. See the ConfigurationOfFuelbaseline19: and look the FuelCompatibilityXXX packages. We could have the same for Squeak if somebody do them. Cheers, Cheers, Max On 11.10.2012, at 11:48, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/10/12 11:04 AM, Max Leske maxle...@gmail.com wrote: Edgar, which version of Squeak are you using? Max Current Squeak 4.4 updated with some code I add. Edgar -- Mariano http://marianopeck.wordpress.com
Re: [Pharo-project] Fuel for Squeak stupid questions
On 10/11/12 12:53 PM, Max Leske maxle...@gmail.com wrote: Ah! Could you try the Date fix I proposed in a 4.3 image? BTW I'm working on fixing the other failing tests (for 4.3) aswell. Cheers, Max You was right, some changed recently in Squeak as 4.3 11860 (fresh unzipped) do not fail the two test of Date, the rest is same as 4.4 You only need fileIn TClassAndTraitDescription from Pharo when Monticello complains Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
That works (at least for me). On 09.10.2012, at 21:59, Mariano Martinez Peck marianop...@gmail.com wrote: On Tue, Oct 9, 2012 at 2:22 PM, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/9/12 4:00 AM, Max Leske maxle...@gmail.com wrote: I guess, your idea was to not use optimized encoding, right? But that would have meant to serialize an extra DateAndTime object. So I hacked together this quick fix (see attachment) that simply encodes another int32 on the stream with the offset. The materialization method took care of creating a new DateAndTime object anyway. With the fix, the Date tests pass. Cheers, Max Max: With your changes the test still fails in Squeak. Very thanks for for show where the problem could be, I try see Date and Timespan implementations In the meanwhile, can you test the workaround I proposed? (that is, remove Date#fuelAccept:) Thanks -- Mariano http://marianopeck.wordpress.com
Re: [Pharo-project] Fuel for Squeak stupid questions
On 10/9/12 4:59 PM, Mariano Martinez Peck marianop...@gmail.com wrote: In the meanwhile, can you test the workaround I proposed? (that is, remove Date#fuelAccept:) Thanks Sorry I miss your previous mail. With Date#fuelAccept: removed we have: '240 run, 233 passes, 2 expected failures, 4 failures, 0 errors, 1 unexpected passes' FLBasicSerializationTest#testGradientFillStyle FLInMemoryBasicSerializationTest#testGradientFillStyle FLMigrationTest#testFormatFixedToVariable FLMigrationTest#testFormatVariableToFixed FLWeakObjectsTest#testAssociationWithWeakFinalizationList Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
Edgar, which version of Squeak are you using? Max On 10.10.2012, at 12:52, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/9/12 4:59 PM, Mariano Martinez Peck marianop...@gmail.com wrote: In the meanwhile, can you test the workaround I proposed? (that is, remove Date#fuelAccept:) Thanks Sorry I miss your previous mail. With Date#fuelAccept: removed we have: '240 run, 233 passes, 2 expected failures, 4 failures, 0 errors, 1 unexpected passes' FLBasicSerializationTest#testGradientFillStyle FLInMemoryBasicSerializationTest#testGradientFillStyle FLMigrationTest#testFormatFixedToVariable FLMigrationTest#testFormatVariableToFixed FLWeakObjectsTest#testAssociationWithWeakFinalizationList Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
I guess, your idea was to not use optimized encoding, right?But that would have meant to serialize an extra DateAndTime object. So I hacked together this quick fix (see attachment) that simply encodes another int32 on the stream with the offset. The materialization method took care of creating a new DateAndTime object anyway.With the fix, the Date tests pass.Cheers,Max Date class-materializeFrom.st Description: Binary data Date-serializeOn.st Description: Binary data On 08.10.2012, at 09:36, Mariano Martinez Peck marianop...@gmail.com wrote:On Mon, Oct 8, 2012 at 8:30 AM, Max Leske maxle...@gmail.com wrote: Looking at the failing tests I see that the objects really are supposed to be different (comparison with #~~). What fails is the equality check (#=) and for Date that is the method Timespan=. The problem that occurs there is a really old (and annoying one), namely that the timezone offset is handled inconsistently: the materialized object has the correct offset while the original object doesn't. I think it could work if you just remove DatefuelAccept: . Can you try? So:- FLBasicSerializationTesttestDate fails because of timezone offset- FLBasicSerializationTesttestGradientFillStyle fails because Object= isn't overridden in any subclass and therefore identity is checked, which fails - (same as above for FLBasicInMemorySerializationTest)The other failures seem to be due to implementation differences between Pharo and Squeak (I didn't really look into them) but I couldn't find one test that seemed to fail because of hashing problems. Yes, I didn't really understand the thing with the hashes. I don't think there should be a problem there. Something obvious is that a normal object that is serialized will probabyt have a different identityHash than the materialized object. Thanks!I'll see if I can come up with a test case. Cheers,MaxOn 07.10.2012, at 16:38, Max Leske maxle...@gmail.com wrote: The two objects I looked at where a string and an OrderedCollection. I wouldn't conclude that the problem is class specific...On 07.10.2012, at 16:12, "Edgar J. De Cleene" edgardec2...@gmail.com wrote: On 10/7/12 10:51 AM, "Max Leske" maxle...@gmail.com wrote: Following you example, in Squeak FLBasicSerializationTest#testDate materialized largeIdentityHash = 2571 anObject. largeIdentityHash = 1094 So why Date serializaded object gives this and the rest of objects no ? Edgar -- Marianohttp://marianopeck.wordpress.com
Re: [Pharo-project] Fuel for Squeak stupid questions
On 10/9/12 4:00 AM, Max Leske maxle...@gmail.com wrote: I guess, your idea was to not use optimized encoding, right? But that would have meant to serialize an extra DateAndTime object. So I hacked together this quick fix (see attachment) that simply encodes another int32 on the stream with the offset. The materialization method took care of creating a new DateAndTime object anyway. With the fix, the Date tests pass. Cheers, Max Max: With your changes the test still fails in Squeak. Very thanks for for show where the problem could be, I try see Date and Timespan implementations Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
Looking at the failing tests I see that the objects really are supposed to be different (comparison with #~~). What fails is the equality check (#=) and for Date that is the method Timespan=. The problem that occurs there is a really old (and annoying one), namely that the timezone offset is handled inconsistently: the materialized object has the correct offset while the original object doesn't. So: - FLBasicSerializationTesttestDate fails because of timezone offset - FLBasicSerializationTesttestGradientFillStyle fails because Object= isn't overridden in any subclass and therefore identity is checked, which fails - (same as above for FLBasicInMemorySerializationTest) The other failures seem to be due to implementation differences between Pharo and Squeak (I didn't really look into them) but I couldn't find one test that seemed to fail because of hashing problems. I'll see if I can come up with a test case. Cheers, Max On 07.10.2012, at 16:38, Max Leske maxle...@gmail.com wrote: The two objects I looked at where a string and an OrderedCollection. I wouldn't conclude that the problem is class specific... On 07.10.2012, at 16:12, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/7/12 10:51 AM, Max Leske maxle...@gmail.com wrote: Following you example, in Squeak FLBasicSerializationTest#testDate materialized largeIdentityHash = 2571 anObject. largeIdentityHash = 1094 So why Date serializaded object gives this and the rest of objects no ? Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
On Mon, Oct 8, 2012 at 8:30 AM, Max Leske maxle...@gmail.com wrote: Looking at the failing tests I see that the objects really are supposed to be different (comparison with #~~). What fails is the equality check (#=) and for Date that is the method Timespan=. The problem that occurs there is a really old (and annoying one), namely that the timezone offset is handled inconsistently: the materialized object has the correct offset while the original object doesn't. I think it could work if you just remove DatefuelAccept: . Can you try? So: - FLBasicSerializationTesttestDate fails because of timezone offset - FLBasicSerializationTesttestGradientFillStyle fails because Object= isn't overridden in any subclass and therefore identity is checked, which fails - (same as above for FLBasicInMemorySerializationTest) The other failures seem to be due to implementation differences between Pharo and Squeak (I didn't really look into them) but I couldn't find one test that seemed to fail because of hashing problems. Yes, I didn't really understand the thing with the hashes. I don't think there should be a problem there. Something obvious is that a normal object that is serialized will probabyt have a different identityHash than the materialized object. Thanks! I'll see if I can come up with a test case. Cheers, Max On 07.10.2012, at 16:38, Max Leske maxle...@gmail.com wrote: The two objects I looked at where a string and an OrderedCollection. I wouldn't conclude that the problem is class specific... On 07.10.2012, at 16:12, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/7/12 10:51 AM, Max Leske maxle...@gmail.com wrote: Following you example, in Squeak FLBasicSerializationTest#testDate materialized largeIdentityHash = 2571 anObject. largeIdentityHash = *1094 * So why Date serializaded object gives this and the rest of objects no ? Edgar -- Mariano http://marianopeck.wordpress.com
[Pharo-project] Fuel for Squeak stupid questions
Hi Mariano: Just have Fuel loaded into Squeak. Trying to understand , first I run Sunit The test says 240 run, 231 passes, 2 expected failures, 6 failures, 0 errors, 1 unexpected passes failures FLBasicSerializationTest#testDate FLBasicSerializationTest#testGradientFillStyle FLInMemoryBasicSerializationTest#testDate FLInMemoryBasicSerializationTest#testGradientFillStyle FLMigrationTest#testFormatFixedToVariable FLMigrationTest#testFormatVariableToFixed FLWeakObjectsTest#testAssociationWithWeakFinalizationList In testDate, clicking in debugger for materialized and for anObject both shows as Date objects with 14 May 2002 as value. Could tell why anObject = materialized evaluates to false ? Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
Here's a possibly related question: after having analysed a large object graph, I get exceptions in FLEncoderencodeReferenceTo:. When I look at the values #largeIdentityHash generates for that object, then the one looked up in the dictionary (that is used by the debugger) is different than the one I get when asking the object manually. Incidentally, the manually retrieved value would lead to a match in the lookup. Could there be something weird going on with the hashes? BTW, the VM is a SqueakVM 4.2.5beta1U. I'll also try with a CogVM later and let you know if it worked. Cheers, Max On 07.10.2012, at 14:38, Edgar J. De Cleene edgardec2...@gmail.com wrote: Hi Mariano: Just have Fuel loaded into Squeak. Trying to understand , first I run Sunit The test says 240 run, 231 passes, 2 expected failures, 6 failures, 0 errors, 1 unexpected passes failures FLBasicSerializationTest#testDate FLBasicSerializationTest#testGradientFillStyle FLInMemoryBasicSerializationTest#testDate FLInMemoryBasicSerializationTest#testGradientFillStyle FLMigrationTest#testFormatFixedToVariable FLMigrationTest#testFormatVariableToFixed FLWeakObjectsTest#testAssociationWithWeakFinalizationList In testDate, clicking in debugger for materialized and for anObject both shows as Date objects with 14 May 2002 as value. Could tell why anObject = materialized evaluates to false ? Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
Hi Edgar. Which version are you using exactly of the package Fuel? Thanks, On Sun, Oct 7, 2012 at 2:38 PM, Edgar J. De Cleene edgardec2...@gmail.comwrote: Hi Mariano: Just have Fuel loaded into Squeak. Trying to understand , first I run Sunit The test says 240 run, 231 passes, 2 expected failures, 6 failures, 0 errors, 1 unexpected passes failures FLBasicSerializationTest#testDate FLBasicSerializationTest#testGradientFillStyle FLInMemoryBasicSerializationTest#testDate FLInMemoryBasicSerializationTest#testGradientFillStyle FLMigrationTest#testFormatFixedToVariable FLMigrationTest#testFormatVariableToFixed FLWeakObjectsTest#testAssociationWithWeakFinalizationList In testDate, clicking in debugger for materialized and for anObject both shows as Date objects with 14 May 2002 as value. Could tell why anObject = materialized evaluates to false ? Edgar -- Mariano http://marianopeck.wordpress.com
Re: [Pharo-project] Fuel for Squeak stupid questions
On 10/7/12 10:16 AM, Max Leske maxle...@gmail.com wrote: Here's a possibly related question: after having analysed a large object graph, I get exceptions in FLEncoderencodeReferenceTo:. When I look at the values #largeIdentityHash generates for that object, then the one looked up in the dictionary (that is used by the debugger) is different than the one I get when asking the object manually. Incidentally, the manually retrieved value would lead to a match in the lookup. Could there be something weird going on with the hashes? BTW, the VM is a SqueakVM 4.2.5beta1U. I'll also try with a CogVM later and let you know if it worked. Cheers, Max Thanks, is a good pointer. In my case I try with last CogVM http://www.mirandabanda.org/files/Cog/VM/VM.r2585/ Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
On 10/7/12 10:23 AM, Mariano Martinez Peck marianop...@gmail.com wrote: Hi Edgar. Which version are you using exactly of the package Fuel? Thanks, From http://ss3.gemstone.com/ss/Fuel repository/ Open and load ConfigurationOfFuel-MarianoMartinezPeck.153 Then ConfigurationOfFuel load And this is the list of loaded from Monticello Browser * ConfigurationOfFuel (MarianoMartinezPeck.153) ³marked dirty as I write ConfigurationOfFuel load in the comment ³ Fuel (MartinDias.685) FuelCompatibilityBeforePharo12 (MartinDias.8) FuelCompatibilityBeforePharo13 (MarianoMartinezPeck.3) FuelCompatibilityBeforePharo14 (MartinDias.3) FuelCompatibilityBeforePharo20 (MartinDias.4) FuelCompatibilityForSqueak (MMP.4) * FuelTests (MartinDias.267) ³marked dirty as I put a self halt for seeing what¹s going on² FuelTestsCompatibilityForSqueak (MMP.1) The same on Pharo-1.4-14557 gives all green test Cheers Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
On 10/7/12 10:51 AM, Max Leske maxle...@gmail.com wrote: Following you example, in Squeak FLBasicSerializationTest#testDate materialized largeIdentityHash = 2571 anObject. largeIdentityHash = 1094 So why Date serializaded object gives this and the rest of objects no ? Edgar
Re: [Pharo-project] Fuel for Squeak stupid questions
The two objects I looked at where a string and an OrderedCollection. I wouldn't conclude that the problem is class specific... On 07.10.2012, at 16:12, Edgar J. De Cleene edgardec2...@gmail.com wrote: On 10/7/12 10:51 AM, Max Leske maxle...@gmail.com wrote: Following you example, in Squeak FLBasicSerializationTest#testDate materialized largeIdentityHash = 2571 anObject. largeIdentityHash = 1094 So why Date serializaded object gives this and the rest of objects no ? Edgar
