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°