[Pharo-project] Experimental Cocoa OS-X based Squeak Cog JIT VM 5.8b7
I've stuck a version (5.8b7) of the cocoa based os-x squeak cog JIT based VM in my experimental folder. Please ensure drawing looks sane, and how it interacts with the mouse and keyboard seems viable. http://homepage.mac.com/johnmci/.Public/experimental/Squeak%205.8b7.app.zip I finished the open/GL tuning and based on benchmarking I simplified the explicit flush required when a Smalltalk Morphic draw cycle fails to do a flush. Lastly I cross check the source code to ensure it compiles correctly for the iOS platform. -- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
Hi John, I tried the VM on Mac OS X 10.5.8 and it just crashes on an image that works perfectly with 5.8.b4. I will soon upgrade to Snow Leopard (finally :)), but I can still play around today if you want me to test further. Cheers, Doru On 3 Sep 2010, at 03:25, John M McIntosh wrote:
I've stuck a version (5.8b7) of the cocoa based os-x squeak cog JIT based VM in my experimental folder. Please ensure drawing looks sane, and how it interacts with the mouse and keyboard seems viable.
http://homepage.mac.com/johnmci/.Public/experimental/Squeak%205.8b7.app.zip
I finished the open/GL tuning and based on benchmarking I simplified the explicit flush required when a Smalltalk Morphic draw cycle fails to do a flush. Lastly I cross check the source code to ensure it compiles correctly for the iOS platform.
-- = = = = = ====================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http:// www.smalltalkconsulting.com = = = = = ======================================================================
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
-- www.tudorgirba.com "Reasonable is what we are accustomed with."
Doru, can you send me the crash report? To fix I'll need that. The VM is compiled with the 10.5.x SDK and should work on it. On 2010-09-02, at 10:25 PM, Tudor Girba wrote:
Hi John,
I tried the VM on Mac OS X 10.5.8 and it just crashes on an image that works perfectly with 5.8.b4.
I will soon upgrade to Snow Leopard (finally :)), but I can still play around today if you want me to test further.
Cheers, Doru
-- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
Hi John, I get no debug.log. The only thing I get is the Apple reporting window saying that it crashed. I attach here the text output of the problem details and system configuration. Cheers, Doru On 3 Sep 2010, at 09:36, John M McIntosh wrote:
Doru, can you send me the crash report? To fix I'll need that. The VM is compiled with the 10.5.x SDK and should work on it.
On 2010-09-02, at 10:25 PM, Tudor Girba wrote:
Hi John,
I tried the VM on Mac OS X 10.5.8 and it just crashes on an image that works perfectly with 5.8.b4.
I will soon upgrade to Snow Leopard (finally :)), but I can still play around today if you want me to test further.
Cheers, Doru
-- = = = = = ====================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http:// www.smalltalkconsulting.com = = = = = ======================================================================
-- www.tudorgirba.com "Obvious things are difficult to teach."
Did a quick try of a fresh Squeak4.2-10382-alpha on your latest 5.8b7, MacPro OS X 10.6.4, NVIDIA GeForce 7300 GT graphics card Resizing the Squeak window still results in graphic artifacts that go away when stopped resizing. Tried 'Update Squeak', got 'Error: unmatched string quote' in scanner while processing update.nice.150.mcm and snapshotting methods. Open Monticello Browser from World menu. +Repository select HTTP, Accept. click 'Open', with http://www.squeaksource.com/ selected. Get 'Error: Could not access http://www.squeaksource.com/' http://www.squeaksource.com/ is accessible from Safari. Ken G. Brown At 6:25 PM -0700 9/2/10, John M McIntosh wrote:
Content-Type: multipart/signed; boundary=Apple-Mail-121-227271873; protocol="application/pkcs7-signature"; micalg=sha1
I've stuck a version (5.8b7) of the cocoa based os-x squeak cog JIT based VM in my experimental folder. Please ensure drawing looks sane, and how it interacts with the mouse and keyboard seems viable.
http://homepage.mac.com/johnmci/.Public/experimental/Squeak%205.8b7.app.zip
I finished the open/GL tuning and based on benchmarking I simplified the explicit flush required when a Smalltalk Morphic draw cycle fails to do a flush. Lastly I cross check the source code to ensure it compiles correctly for the iOS platform.
-- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
Attachment converted: MacProHD0:smime 764.p7s ( / ) (079FC4B2)
John M McIntosh wrote:
I've stuck a version (5.8b7) of the cocoa based os-x squeak cog JIT based VM in my experimental folder. Please ensure drawing looks sane, and how it interacts with the mouse and keyboard seems viable.
http://homepage.mac.com/johnmci/.Public/experimental/Squeak%205.8b7.app.zip
I finished the open/GL tuning and based on benchmarking I simplified the explicit flush required when a Smalltalk Morphic draw cycle fails to do a flush. Lastly I cross check the source code to ensure it compiles correctly for the iOS platform.
-- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ==========================================================================
Hi John, I found something strange with reported file creation and modification time. The values answered by #primLookupEntryIn:index: are (Tested on Squeak 4.1, Mac OS 10.5.8): 4.2.4 VM (Stack VM): 3461318674 3461318674 newFile 3449660476 3447966888 Squeak 4.2.4beta1U.app 3460975777 3460916669 Squeak.app 3460551046 3460551046 Squeak4.1.image 3449660435 3448977726 SqueakV41.sources 5.8b7 (Cog experimental VM) 3461314066 3461314066 newFile 3447962280 3447962280 Squeak 4.2.4beta1U.app 3460912061 3460912061 Squeak.app 3448973124 3460546438 Squeak4.1.image 3448973118 3448973118 SqueakV41.sources Take for instance SqueakV41.sources: DateAndTime fromSeconds: 3449660435 2010-04-25T15:00:35-03:00 DateAndTime fromSeconds: 3448977726 2010-04-17T17:22:06-03:00 "This is the correct one according to Finder, 4.2.4 modification time" DateAndTime fromSeconds: 3448973118 2010-04-17T16:05:18-03:00 DateAndTime fromSeconds: 3448973118 2010-04-17T16:05:18-03:00 So I'm, getting incorrect time info in the FileList. Something else I see is that 4.2.4 VM answers the same value for creation and modification time for some files, while 5.8b7 answers the same value for other files. (Why?) For the newly created "newFile", both answer the same value for creation and modification time. The correct value is (again) the one answered by 4.2.4. I'm at the ART time zone. Thanks, Juan Vuletich
Ok, well it's great someone is actually paying attention. In 4.2.4 if (stat(unixPath, &statBuf) && lstat(unixPath, &statBuf)) and pull these values time_t st_mtime; /* [XSI] Last data modification time */ time_t st_ctime; /* [XSI] Time of last status change */ For your examples the lstat() is the one populating the structure. Also described in the man page as the time file was last accessed or modified, of when the inode was last changed, or the birth time of the inode. We then consider localtime(&unixTime)->tm_gmtoff; to create a time in the Squeak epoch. Now for the Cocoa Version we take NSFileCreationDate The key in a file attribute dictionary whose value indicates the file's creation date. NSFileModificationDate The key in a file attribute dictionary whose value indicates the file's last modified date. { NSFileCreationDate = 2010-08-20 13:25:54 -0700; NSFileExtendedAttributes = { "com.apple.quarantine" = <30303030 3b346337 36396138 623b5361 66617269 2e617070 3b7c636f 6d2e6170 706c652e 53616661 7269>; }; NSFileExtensionHidden = 0; NSFileGroupOwnerAccountID = 0; NSFileGroupOwnerAccountName = wheel; NSFileHFSCreatorCode = 1178686292; NSFileHFSTypeCode = 1398039400; NSFileModificationDate = 2010-09-06 18:00:05 -0700; NSFileOwnerAccountID = 501; NSFileOwnerAccountName = johnmci; NSFilePosixPermissions = 420; NSFileReferenceCount = 1; NSFileSize = 4073312; NSFileSystemFileNumber = 81731241; NSFileSystemNumber = 234881026; NSFileType = NSFileTypeRegular; } Now actually another bug is that I'm not re-calculating for DayLightSavings time. I've fixed that for 5.8b10 where the date below shows as 17:00:05 but really it should be 18:00:05 In checking the creation time in Squeak I see it thinks it is "2010-08-20T13:25:54+00:00" which baring the timezone/daylightsavings +00.00? issue seems correctly stated. Let me send you a 5.8b10 VM to test /Users/johnmci/Shared/TextFlow-trunk/trunk.changes On 2010-09-07, at 9:53 AM, Juan Vuletich wrote:
John M McIntosh wrote:
I've stuck a version (5.8b7) of the cocoa based os-x squeak cog JIT based VM in my experimental folder. Please ensure drawing looks sane, and how it interacts with the mouse and keyboard seems viable. http://homepage.mac.com/johnmci/.Public/experimental/Squeak%205.8b7.app.zip
I finished the open/GL tuning and based on benchmarking I simplified the explicit flush required when a Smalltalk Morphic draw cycle fails to do a flush. Lastly I cross check the source code to ensure it compiles correctly for the iOS platform. -- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ==========================================================================
Hi John,
I found something strange with reported file creation and modification time. The values answered by #primLookupEntryIn:index: are (Tested on Squeak 4.1, Mac OS 10.5.8):
4.2.4 VM (Stack VM): 3461318674 3461318674 newFile 3449660476 3447966888 Squeak 4.2.4beta1U.app 3460975777 3460916669 Squeak.app 3460551046 3460551046 Squeak4.1.image 3449660435 3448977726 SqueakV41.sources
5.8b7 (Cog experimental VM) 3461314066 3461314066 newFile 3447962280 3447962280 Squeak 4.2.4beta1U.app 3460912061 3460912061 Squeak.app 3448973124 3460546438 Squeak4.1.image 3448973118 3448973118 SqueakV41.sources
Take for instance SqueakV41.sources: DateAndTime fromSeconds: 3449660435 2010-04-25T15:00:35-03:00 DateAndTime fromSeconds: 3448977726 2010-04-17T17:22:06-03:00 "This is the correct one according to Finder, 4.2.4 modification time" DateAndTime fromSeconds: 3448973118 2010-04-17T16:05:18-03:00 DateAndTime fromSeconds: 3448973118 2010-04-17T16:05:18-03:00
So I'm, getting incorrect time info in the FileList.
Something else I see is that 4.2.4 VM answers the same value for creation and modification time for some files, while 5.8b7 answers the same value for other files. (Why?) For the newly created "newFile", both answer the same value for creation and modification time. The correct value is (again) the one answered by 4.2.4.
I'm at the ART time zone.
Thanks, Juan Vuletich
-- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
John M McIntosh wrote:
Ok, well it's great someone is actually paying attention.
In 4.2.4 if (stat(unixPath, &statBuf) && lstat(unixPath, &statBuf))
and pull these values
time_t st_mtime; /* [XSI] Last data modification time */ time_t st_ctime; /* [XSI] Time of last status change */
For your examples the lstat() is the one populating the structure.
Also described in the man page as the time file was last accessed or modified, of when the inode was last changed, or the birth time of the inode.
We then consider localtime(&unixTime)->tm_gmtoff;
to create a time in the Squeak epoch.
Now for the Cocoa Version we take NSFileCreationDate The key in a file attribute dictionary whose value indicates the file's creation date. NSFileModificationDate The key in a file attribute dictionary whose value indicates the file's last modified date.
{ NSFileCreationDate = 2010-08-20 13:25:54 -0700; NSFileExtendedAttributes = { "com.apple.quarantine" = <30303030 3b346337 36396138 623b5361 66617269 2e617070 3b7c636f 6d2e6170 706c652e 53616661 7269>; }; NSFileExtensionHidden = 0; NSFileGroupOwnerAccountID = 0; NSFileGroupOwnerAccountName = wheel; NSFileHFSCreatorCode = 1178686292; NSFileHFSTypeCode = 1398039400; NSFileModificationDate = 2010-09-06 18:00:05 -0700; NSFileOwnerAccountID = 501; NSFileOwnerAccountName = johnmci; NSFilePosixPermissions = 420; NSFileReferenceCount = 1; NSFileSize = 4073312; NSFileSystemFileNumber = 81731241; NSFileSystemNumber = 234881026; NSFileType = NSFileTypeRegular; }
Now actually another bug is that I'm not re-calculating for DayLightSavings time.
I've fixed that for 5.8b10 where the date below shows as 17:00:05 but really it should be 18:00:05
In checking the creation time in Squeak I see it thinks it is "2010-08-20T13:25:54+00:00" which baring the timezone/daylightsavings +00.00? issue seems correctly stated.
Let me send you a 5.8b10 VM to test
/Users/johnmci/Shared/TextFlow-trunk/trunk.changes
------------------------------------------------------------------------
On 2010-09-07, at 9:53 AM, Juan Vuletich wrote:
John M McIntosh wrote:
I've stuck a version (5.8b7) of the cocoa based os-x squeak cog JIT based VM in my experimental folder. Please ensure drawing looks sane, and how it interacts with the mouse and keyboard seems viable. http://homepage.mac.com/johnmci/.Public/experimental/Squeak%205.8b7.app.zip
I finished the open/GL tuning and based on benchmarking I simplified the explicit flush required when a Smalltalk Morphic draw cycle fails to do a flush. Lastly I cross check the source code to ensure it compiles correctly for the iOS platform. -- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ==========================================================================
Hi John,
I found something strange with reported file creation and modification time. The values answered by #primLookupEntryIn:index: are (Tested on Squeak 4.1, Mac OS 10.5.8):
4.2.4 VM (Stack VM): 3461318674 3461318674 newFile 3449660476 3447966888 Squeak 4.2.4beta1U.app 3460975777 3460916669 Squeak.app 3460551046 3460551046 Squeak4.1.image 3449660435 3448977726 SqueakV41.sources
5.8b7 (Cog experimental VM) 3461314066 3461314066 newFile 3447962280 3447962280 Squeak 4.2.4beta1U.app 3460912061 3460912061 Squeak.app 3448973124 3460546438 Squeak4.1.image 3448973118 3448973118 SqueakV41.sources
Take for instance SqueakV41.sources: DateAndTime fromSeconds: 3449660435 2010-04-25T15:00:35-03:00 DateAndTime fromSeconds: 3448977726 2010-04-17T17:22:06-03:00 "This is the correct one according to Finder, 4.2.4 modification time" DateAndTime fromSeconds: 3448973118 2010-04-17T16:05:18-03:00 DateAndTime fromSeconds: 3448973118 2010-04-17T16:05:18-03:00
So I'm, getting incorrect time info in the FileList.
Something else I see is that 4.2.4 VM answers the same value for creation and modification time for some files, while 5.8b7 answers the same value for other files. (Why?) For the newly created "newFile", both answer the same value for creation and modification time. The correct value is (again) the one answered by 4.2.4.
I'm at the ART time zone.
Thanks, Juan Vuletich
-- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
Hi John, I've just tested the new VM you sent me. The details are below. This is what I see: 1)- 5.8b10 gives exactly the same date/times as 5.8b7 2)- 5.8 (both) seem to show times that are about 1:17 (one hour, seventeen minutes) earlier than the correct one. This doesn't have relation with time zones or daylight savings, it is a constant number of seconds, but not entire hours. May be it is related to the Mac OS version? I'm using MacOS 10.5.8. 3)- 5.8 (both) agree with Macintosh Explorer on when creation time should equal modification time. This is good. I believe that the new API is the correct one, even if still off by 01:17. 4)- 4.2.4 seems to know my time zone (-03:00), 5.8 (both) seem not to know. 5)- Creation time for 4.2.4 seems just wrong. Some times it is the same as modification time (when it shouldnt). Some times it doesn't make sense to me. I believe you should see similar results on your system, right? Thanks, Juan Vuletich Creation time left. Modification time right. 5.8b7 2010-09-07 T 12:07:46 +00:00 2010-09-07 T 12:07:46 +00:00 popo 2010-04-05 T 23:18:00 +00:00 2010-04-05 T 23:18:00 +00:00 Squeak 4.2.4beta1U.app 2010-09-07 T 16:41:29 +00:00 2010-09-07 T 16:41:29 +00:00 Squeak 5.8b10.app 2010-01-25 T 15:49:27 +00:00 2010-01-25 T 15:49:27 +00:00 Squeak 64-32 5.4b2.app 2010-09-02 T 20:27:41 +00:00 2010-09-02 T 20:27:41 +00:00 Squeak.app 2010-04-17 T 16:05:24 +00:00 2010-08-29 T 14:53:58 +00:00 Squeak4.1.image 2010-04-25 T 13:42:47 +00:00 2010-04-25 T 13:42:48 +00:00 SqueakV40.sources 2010-04-17 T 16:05:18 +00:00 2010-04-17 T 16:05:18 +00:00 SqueakV41.sources 5.8b10 2010-09-07 T 12:07:46 +00:00 2010-09-07 T 12:07:46 +00:00 popo 2010-04-05 T 23:18:00 +00:00 2010-04-05 T 23:18:00 +00:00 Squeak 4.2.4beta1U.app 2010-09-07 T 16:41:29 +00:00 2010-09-07 T 16:41:29 +00:00 Squeak 5.8b10.app 2010-01-25 T 15:49:27 +00:00 2010-01-25 T 15:49:27 +00:00 Squeak 64-32 5.4b2.app 2010-09-02 T 20:27:41 +00:00 2010-09-02 T 20:27:41 +00:00 Squeak.app 2010-04-17 T 16:05:24 +00:00 2010-08-29 T 14:53:58 +00:00 Squeak4.1.image 2010-04-25 T 13:42:47 +00:00 2010-04-25 T 13:42:48 +00:00 SqueakV40.sources 2010-04-17 T 16:05:18 +00:00 2010-04-17 T 16:05:18 +00:00 SqueakV41.sources 4.2.4 2010-09-08 T 08:12:29 -03:00 2010-09-07 T 13:24:34 -03:00 popo 2010-04-25 T 15:01:16 -03:00 2010-04-06 T 00:34:48 -03:00 Squeak 4.2.4beta1U.app 2010-09-08 T 08:14:53 -03:00 2010-09-07 T 17:58:17 -03:00 Squeak 5.8b10.app 2010-09-08 T 08:09:27 -03:00 2010-01-25 T 18:06:15 -03:00 Squeak 64-32 5.4b2.app 2010-09-03 T 14:09:37 -03:00 2010-09-02 T 21:44:29 -03:00 Squeak.app 2010-09-08 T 08:14:18 -03:00 2010-08-29 T 16:10:46 -03:00 Squeak4.1.image 2010-04-25 T 14:59:36 -03:00 2010-04-25 T 14:59:36 -03:00 SqueakV40.sources 2010-09-07 T 13:37:40 -03:00 2010-04-17 T 17:22:06 -03:00 SqueakV41.sources Rage Macintosh Explorer 2010-09-07 13:24:34 2010-09-07 13:24:34 popo 2010-04-06 12:34:48 2010-04-06 12:34:48 Squeak 4.2.4beta1U.app 2010-09-07 17:58:17 2010-09-07 17:58:17 Squeak 5.8b10.app 2010-01-25 18:06:15 2010-01-25 18:06:15 Squeak 64-32 5.4b2.app 2010-09-02 21:44:29 2010-09-02 21:44:29 Squeak.app 2010-04-17 17:22:12 2010-08-29 16:10:46 Squeak4.1.image 2010-04-25 14:59:35 2010-04-25 14:59:36 SqueakV40.sources 2010-04-17 17:22:06 2010-04-17 17:02:06 SqueakV41.sources
Hi John,
I've just tested the new VM you sent me. The details are below. This is what I see:
1)- 5.8b10 gives exactly the same date/times as 5.8b7
2)- 5.8 (both) seem to show times that are about 1:17 (one hour, seventeen minutes) earlier than the correct one. This doesn't have relation with time zones or daylight savings, it is a constant number of seconds, but not entire hours. May be it is related to the Mac OS version? I'm using MacOS 10.5.8.
3)- 5.8 (both) agree with Macintosh Explorer on when creation time should equal modification time. This is good. I believe that the new API is the correct one, even if still off by 01:17.
4)- 4.2.4 seems to know my time zone (-03:00), 5.8 (both) seem not to know.
5)- Creation time for 4.2.4 seems just wrong. Some times it is the same as modification time (when it shouldnt). Some times it doesn't make sense to me.
I believe you should see similar results on your system, right?
Thanks, Juan Vuletich
Well I think I know what the issue is, Juan can you tell me what timezone and city you have set in the "Date & Time" TimeZone tab in system settings? -- =========================================================================== John M. McIntosh <johnmci@smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================
participants (4)
-
John M McIntosh -
Juan Vuletich -
Ken G. Brown -
Tudor Girba