Hello all,
I bailed on getting NaCl working, though the support for libsodium FFI is there. However, I was able to complete my long efforts at moving ASN1 to recognize class tagging. And I have a test! Thus these changes to Crypto's ASN1 is to support my efforts with EParrotTalk and I believe I am finished with these changes. #GREEN! Andso, I will put a seal on this release of ProCrypto.release.3-rabbt.60.mcm.
Nota bene: I am automatically updating from the alpha truck every time the image resumes.
Installer ss project: 'Cryptography'; install: 'ProCrypto.release.3'.
In ParrotTalk, I got all the v39 headers working, Now to the protocol state machine, though they are still being marshaled with a choice, so I am not yet directly marshaling them as first class ASN1Types with the applicationClass class tag <2r01>. I am using numeric tags 0-31 between ParrotTalk and PromisesFar. I have registering them all, so I am ready to remove the choice. You fail 100% of the time you don't take your shot.
Cheers!
-- â¤ď¸âđĽđ°
Alright, I removed the choice from ParrotTalk. I would share a note about Promises. I removed the PriorityVat and am now scheduling eventual sends on the ProcesssorScheduler, which maintains ordering. This allows the use of Semaphore waits. Performs better too.
Good morning, rabbit
On 7/4/23 23:11, rabbit wrote:
Hello all,
I bailed on getting NaCl working, though the support for libsodium FFI is there. However, I was able to complete my long efforts at moving ASN1 to recognize class tagging. And I have a test! Thus these changes to Crypto's ASN1 is to support my efforts with EParrotTalk and I believe I am finished with these changes. #GREEN! Andso, I will put a seal on this release of ProCrypto.release.3-rabbt.60.mcm.
Nota bene: I am automatically updating from the alpha truck every time the image resumes.
Installer ss project: 'Cryptography'; install: 'ProCrypto.release.3'.
In ParrotTalk, I got all the v39 headers working, Now to the protocol state machine, though they are still being marshaled with a choice, so I am not yet directly marshaling them as first class ASN1Types with the applicationClass class tag <2r01>. I am using numeric tags 0-31 between ParrotTalk and PromisesFar. I have registering them all, so I am ready to remove the choice. You fail 100% of the time you don't take your shot.
Cheers!
-- â¤ď¸âđĽđ°
-- â¤ď¸âđĽđ°
I would be clear with today's updates to ProCrypto.release.3.
- I folded code into Crypto, and brought in TraceMonitor, Registers (slight mods to init), PromisesLocal and ELinda. and so it is all down to one package now, the erstwhile Crypto-rabbt.16.mcz. However, for the release I continue to specifying ProCrypto.release.3 as the release artifact. The locks down version 16 of Crypto, so it's not flying in the wind again.Sorry about that. Bad bunny! - I had to add an accessor for ClassForPlatform class var, in both RGRegister superclasses to call initialize and provide a valid class for Register instantiation. There was an #initialize load order issue as MD5 wanted to initialize first. - All test pass, obviously. - Now using a SchedulerVat in Promises so the send activation Process have the same priority as the sending Process. I may have a #schedule:priority: on the Vat but it is not exposed to the send site. I'm thinking of getting to QoS through thisContext. Is there a way to store a link there or use a Process local var to access ParrotTalkQoS for sending? Perhaps specify QoS when creating the eventual reference in the first place, something like <#eventualPriority: EParrotTalk flashPriority>. - ASN1:
- I am opening a button bar to inspect aspects of
- ASN1 Modules - Type Registrations by ASN1Tag - Type Initializers I am using to control load order. - If something is defined that's missing in an import, the TypeReference, which is a hated class of mine (I wrote it. You know this feeling?), is installed in the asn1Module of the type being defined, not in the proper import it's supposed to come from. and so when the import gets defined, those TypeReferences are not #becomeForward: -ed to the proper type construction finally being loaded. So it fails. a touch tricky. - Fails in a - Learned of the Preference for update on image start, so I'm using that in my local space so I don't impose on y'all.
Enjoy and hit me up with your thoughts. Appreciate.
â¤ď¸âđĽđ°
On 7/5/23 08:13, rabbit wrote:
Alright, I removed the choice from ParrotTalk. I would share a note about Promises. I removed the PriorityVat and am now scheduling eventual sends on the ProcesssorScheduler, which maintains ordering. This allows the use of Semaphore waits. Performs better too.
Good morning, rabbit
On 7/4/23 23:11, rabbit wrote:
Hello all,
I bailed on getting NaCl working, though the support for libsodium FFI is there. However, I was able to complete my long efforts at moving ASN1 to recognize class tagging. And I have a test! Thus these changes to Crypto's ASN1 is to support my efforts with EParrotTalk and I believe I am finished with these changes. #GREEN! Andso, I will put a seal on this release of ProCrypto.release.3-rabbt.60.mcm.
Nota bene: I am automatically updating from the alpha truck every time the image resumes.
Installer ss project: 'Cryptography'; install: 'ProCrypto.release.3'.
In ParrotTalk, I got all the v39 headers working, Now to the protocol state machine, though they are still being marshaled with a choice, so I am not yet directly marshaling them as first class ASN1Types with the applicationClass class tag <2r01>. I am using numeric tags 0-31 between ParrotTalk and PromisesFar. I have registering them all, so I am ready to remove the choice. You fail 100% of the time you don't take your shot.
Cheers!
-- â¤ď¸âđĽđ°
-- â¤ď¸âđĽđ°
-- â¤ď¸âđĽđ°
Ahh, always yet another...
- Added the bones for ESharedMemory eventual refences, copying the classes from EFarPromes but left out encoder / cipher serializer stuff. not the something... - a skeleton, not interested in pursuing that rabbit hole. I am focused on building upo and getting EFarPromises working. - my plans are for next next bring forth replica sessions, with 3 concurrent, individually secured sessions wired to the same remote object.
- 3 replica sessions to a single destination - each connection is indirect, going through peer Bridges, including your own you registered with your Broker Quorum. - First come, first serve messaging. - ObjectSerialization occurs at the composite session stack, encryption in each crypt sessions. - Quorum among the Three Cloud Sisters stirring the Cauldron and synchronizing a teatime clock. Monitor messages exchanged with their peer Brokers, updates this time, and the Brokers pass time to the Bridges. - Questioning how to establish initial Broker hookup (UDP? Bluetooth mesh network to the Internet?) and ensure a Certificate issued by CAs of CallistoHouse. - Cache-pushing of the Naming service to the Brokers. - 2 ways to get to a service, Broker discovery through naming and introduction through activity. - Cipher rendezvous cycling - Bridge cycling - Broker cycling - The objective? Completely and totally impervious service networking.
On 7/5/23 21:58, rabbit wrote:
I would be clear with today's updates to ProCrypto.release.3.
I folded code into Crypto, and brought in TraceMonitor, Registers (slight mods to init), PromisesLocal and ELinda. and so it is all down to one package now, the erstwhile Crypto-rabbt.16.mcz. However, for the release I continue to specifying ProCrypto.release.3 as the release artifact. The locks down version 16 of Crypto, so it's not flying in the wind again.Sorry about that. Bad bunny!
I had to add an accessor for ClassForPlatform class var, in both RGRegister superclasses to call initialize and provide a valid class for Register instantiation. There was an #initialize load order issue as MD5 wanted to initialize first.
All test pass, obviously.
Now using a SchedulerVat in Promises so the send activation Process have the same priority as the sending Process. I may have a #schedule:priority: on the Vat but it is not exposed to the send site. I'm thinking of getting to QoS through thisContext. Is there a way to store a link there or use a Process local var to access ParrotTalkQoS for sending? Perhaps specify QoS when creating the eventual reference in the first place, something like <#eventualPriority: EParrotTalk flashPriority>.
ASN1:
I am opening a button bar to inspect aspects of
ASN1 Modules
Type Registrations by ASN1Tag
Type Initializers I am using to control load order.
If something is defined that's missing in an import, the TypeReference, which is a hated class of mine (I wrote it. You know this feeling?), is installed in the asn1Module of the type being defined, not in the proper import it's supposed to come from. and so when the import gets defined, those TypeReferences are not #becomeForward: -ed to the proper type construction finally being loaded. So it fails. a touch tricky.
Fails in a
Learned of the Preference for update on image start, so I'm using that in my local space so I don't impose on y'all.
Enjoy and hit me up with your thoughts. Appreciate.
â¤ď¸âđĽđ°
On 7/5/23 08:13, rabbit wrote:
Alright, I removed the choice from ParrotTalk. I would share a note about Promises. I removed the PriorityVat and am now scheduling eventual sends on the ProcesssorScheduler, which maintains ordering. This allows the use of Semaphore waits. Performs better too.
Good morning, rabbit
On 7/4/23 23:11, rabbit wrote:
Hello all,
I bailed on getting NaCl working, though the support for libsodium FFI is there. However, I was able to complete my long efforts at moving ASN1 to recognize class tagging. And I have a test! Thus these changes to Crypto's ASN1 is to support my efforts with EParrotTalk and I believe I am finished with these changes. #GREEN! Andso, I will put a seal on this release of ProCrypto.release.3-rabbt.60.mcm.
Nota bene: I am automatically updating from the alpha truck every time the image resumes.
Installer ss project: 'Cryptography'; install: 'ProCrypto.release.3'.
In ParrotTalk, I got all the v39 headers working, Now to the protocol state machine, though they are still being marshaled with a choice, so I am not yet directly marshaling them as first class ASN1Types with the applicationClass class tag <2r01>. I am using numeric tags 0-31 between ParrotTalk and PromisesFar. I have registering them all, so I am ready to remove the choice. You fail 100% of the time you don't take your shot.
Cheers!
-- â¤ď¸âđĽđ°
-- â¤ď¸âđĽđ°
-- â¤ď¸âđĽđ°
-- â¤ď¸âđĽđ°
On 7/5/23 21:58, rabbit wrote:
ASN1:
I am opening a button bar to inspect aspects of
ASN1 Modules
Type Registrations by ASN1Tag
Type Initializers I am using to control load order.
If something is defined that's missing in an import, the TypeReference, which is a hated class of mine (I wrote it. You know this feeling?), is installed in the asn1Module of the type being defined, not in the proper import it's supposed to come from. and so when the import gets defined, those TypeReferences are not #becomeForward: -ed to the proper type construction finally being loaded. So it fails. a touch tricky.
Fails in a
- As there is a difference between
- a type registration under an ASN1Tag <class ! numeric ! isConstructed> in the ASN1Module ASN1ClassToTagToTypeDispatchTable - and also an ASN1MappedSequenceType defined with a Module. - Lookup starts with the DispatchTable with ASN1Tag of the desired type, followed by a search through the local definitions and then scans the imports for the Named Type - So we really have:
- ASN1TaggedTypes <class ! numeric ! isConstructed> - NamedTypes <moduleName ! typeName> - So, as I am processing ParrotTalk traffic and pushing and popping inner headers out of the wrapper headers, all types are en/decoded through the DispatchTable, by ASN1Type. - In order to actually fully marshal the data, I lookup the defined SequenceType by Name and use the ASN1BasicTypes to hydrate/desicate.
- Learned of the Preference for update on image start, so I'm using that in my local space so I don't impose on y'all.
Enjoy and hit me up with your thoughts. Appreciate.
Tak. Arrivederch, rabbit.
â¤ď¸âđĽđ°
On 7/5/23 08:13, rabbit wrote:
Alright, I removed the choice from ParrotTalk. I would share a note about Promises. I removed the PriorityVat and am now scheduling eventual sends on the ProcesssorScheduler, which maintains ordering. This allows the use of Semaphore waits. Performs better too.
Good morning, rabbit
On 7/4/23 23:11, rabbit wrote:
Hello all,
I bailed on getting NaCl working, though the support for libsodium FFI is there. However, I was able to complete my long efforts at moving ASN1 to recognize class tagging. And I have a test! Thus these changes to Crypto's ASN1 is to support my efforts with EParrotTalk and I believe I am finished with these changes. #GREEN! Andso, I will put a seal on this release of ProCrypto.release.3-rabbt.60.mcm.
Nota bene: I am automatically updating from the alpha truck every time the image resumes.
Installer ss project: 'Cryptography'; install: 'ProCrypto.release.3'.
In ParrotTalk, I got all the v39 headers working, Now to the protocol state machine, though they are still being marshaled with a choice, so I am not yet directly marshaling them as first class ASN1Types with the applicationClass class tag <2r01>. I am using numeric tags 0-31 between ParrotTalk and PromisesFar. I have registering them all, so I am ready to remove the choice. You fail 100% of the time you don't take your shot.
Cheers!
-- â¤ď¸âđĽđ°
-- â¤ď¸âđĽđ°
-- â¤ď¸âđĽđ°
-- â¤ď¸âđĽđ°
squeak-dev@lists.squeakfoundation.org
-
rabbit