Are you sure this is a safe thing to do? My understanding is Monticello
treats
the two-part category names as:
PackageName-Category name
Not if you start adding packages that are named 'Package-Name' ;)
From this I gather your answer is "yes" this is what you want to do. I suppose
you have a lot more confidence in the tools than I do. I think we will get bit by this someday, but I have nevertheless done what you want. We now have all separate algorithm packages. However, I decided against Monticello prereqs; one change to Base and you suddenly have to resave *eleven* packages, and we all know how buggy prereqs are. But I'll go any direction the team wants to on this..
I follwed hmm's lead on removing ASN1Constants pool dictionary, it was causing a load warning, but not sure whether it was needed or not. I know nothing about ASN1. Further, I noticed some of the base-class extensions named "decodeAsn1Der" were also used by DSA. Therefore, ALL ASN1 extensions are in the CryptographyBase package. Someone who knows about ASN1 should clean this up.
Remember, you need to load CryptographBase-cmm.3.mcz, NOT version 5. Also, Cryptography-Support is not needed.
After you load Base.3, then load each algorithm-specific package and, finally, the Tests package. 15 of 18 of the tests pass, just as if I load Cryptography 0.3 from SqueakMap.
- Chris
On 10/24/05, Chris Muller chris@funkyobjects.org wrote:
I suppose you have a lot more confidence in the tools than I do.
Yup :)
However, I decided against Monticello prereqs; one change to Base and you suddenly have to resave *eleven* packages, and we all know how buggy prereqs are.
You may want to try a diamond shape: Cryptography-All (dummy package) Cryptography-MD4 Cryptography-Base Cryptography-MD5 ...
and agree with the team that you always save on the Cryptography-All level. That's what we do with Kolibri, and it works fine.
But formally I'm not even a member of the team, so what am I doing here ;)
KISS is probably the way to go for now anyway.
Thanks Chris and Cees for working through this. I wish I had more to offer. I would think that most people that would need cryptology would not need all packages. It's probably better to leave them as separate packages. The Base package is needed to eliminate duplication of those classes that are used by all.
Maybe Hans-Martin or Rob have more to offer on the structure?
I'm still working through ASN.1.
Ron
-----Original Message----- From: cryptography-bounces@lists.squeakfoundation.org [mailto:cryptography-bounces@lists.squeakfoundation.org] On Behalf Of Cees De Groot Sent: Monday, October 24, 2005 1:17 PM To: chris@funkyobjects.org; Cryptography Team Development List Subject: Re: [Cryptography Team] Initial package factoring complete (was: Re:some more repository things)
On 10/24/05, Chris Muller chris@funkyobjects.org wrote:
I suppose you have a lot more confidence in the tools than I do.
Yup :)
However, I decided against Monticello prereqs; one change to Base and you suddenly have to resave *eleven* packages, and
we
all know how buggy prereqs are.
You may want to try a diamond shape: Cryptography-All (dummy package) Cryptography-MD4 Cryptography-Base Cryptography-MD5 ...
and agree with the team that you always save on the Cryptography-All level. That's what we do with Kolibri, and it works fine.
But formally I'm not even a member of the team, so what am I doing here ;)
KISS is probably the way to go for now anyway. _______________________________________________ Cryptography mailing list Cryptography@lists.squeakfoundation.org http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/cryptography
Ron Teitelbaum wrote:
Maybe Hans-Martin or Rob have more to offer on the structure?
I like Chris' structure and am currently loading it into my image. We're free to play around until we find something that works well for both developers and end users. At the moment, I'd put more weight on developer comfort so that we get work done. If we then find that the structure is too cuimbersome for end users, we can still change it.
Cheers, Hans-Martin
On 10/24/05, Hans-Martin Mosner hmm@heeg.de wrote:
At the moment, I'd put more weight on developer comfort so that we get work done. If we then find that the structure is too cuimbersome for end users, we can still change it.
Very wise words, Herr Doktor Mosner...
Chris,
Is there a way to push your version 3 to a version 6 to remove ambiguity? I loaded my changes to make SHA256 work, including a change to ThirtyTwoBitRegister I included that is a category *Cryptography-SHA256-..." Was that what you were intending or should that additional change have gone somewhere else? Should we delete the support package?
Hans-Martin, Also I fixed errors in the test package, so now only the DES primitive error is left. Were you able to work on that, how's it going?
All, I'm about half way through the ASN.1 documentation. My thoughts so far are to attempt a full implementation because there are other pieces of this that I need on my own project.
Ron
-----Original Message----- From: cryptography-bounces@lists.squeakfoundation.org [mailto:cryptography-bounces@lists.squeakfoundation.org] On Behalf Of Chris Muller Sent: Monday, October 24, 2005 12:41 PM To: cryptography@lists.squeakfoundation.org Subject: [Cryptography Team] Initial package factoring complete (was: Re:some more repository things)
Are you sure this is a safe thing to do? My understanding is Monticello
treats
the two-part category names as:
PackageName-Category name
Not if you start adding packages that are named 'Package-Name' ;)
From this I gather your answer is "yes" this is what you want to do. I
suppose you have a lot more confidence in the tools than I do. I think we will get bit by this someday, but I have nevertheless done what you want. We now have all separate algorithm packages. However, I decided against Monticello prereqs; one change to Base and you suddenly have to resave *eleven* packages, and we all know how buggy prereqs are. But I'll go any direction the team wants to on this..
I follwed hmm's lead on removing ASN1Constants pool dictionary, it was causing a load warning, but not sure whether it was needed or not. I know nothing about ASN1. Further, I noticed some of the base-class extensions named "decodeAsn1Der" were also used by DSA. Therefore, ALL ASN1 extensions are in the CryptographyBase package. Someone who knows about ASN1 should clean this up.
Remember, you need to load CryptographBase-cmm.3.mcz, NOT version 5. Also, Cryptography-Support is not needed.
After you load Base.3, then load each algorithm-specific package and, finally, the Tests package. 15 of 18 of the tests pass, just as if I load Cryptography 0.3 from SqueakMap.
- Chris _______________________________________________ Cryptography mailing list Cryptography@lists.squeakfoundation.org http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/cryptography
I answered my own question about ThirtyTwoBitRegister, and moved it to Cryptography-Core
Ron
-----Original Message----- From: cryptography-bounces@lists.squeakfoundation.org [mailto:cryptography-bounces@lists.squeakfoundation.org] On Behalf Of Ron Teitelbaum Sent: Friday, October 28, 2005 3:22 PM To: 'Cryptography Team Development List' Subject: RE: [Cryptography Team] Initial package factoring complete (was:Re:some more repository things)
Chris,
Is there a way to push your version 3 to a version 6 to remove ambiguity? I loaded my changes to make SHA256 work, including a change to ThirtyTwoBitRegister I included that is a category *Cryptography-SHA256-..." Was that what you were intending or should that additional change have gone somewhere else? Should we delete the support package?
Hans-Martin, Also I fixed errors in the test package, so now only the DES primitive error is left. Were you able to work on that, how's it going?
All, I'm about half way through the ASN.1 documentation. My thoughts so far are to attempt a full implementation because there are other pieces of this that I need on my own project.
Ron
-----Original Message----- From: cryptography-bounces@lists.squeakfoundation.org [mailto:cryptography-bounces@lists.squeakfoundation.org] On Behalf Of Chris Muller Sent: Monday, October 24, 2005 12:41 PM To: cryptography@lists.squeakfoundation.org Subject: [Cryptography Team] Initial package factoring complete (was: Re:some more repository things)
Are you sure this is a safe thing to do? My understanding is Monticello
treats
the two-part category names as:
PackageName-Category name
Not if you start adding packages that are named 'Package-Name' ;)
From this I gather your answer is "yes" this is what you want to do. I
suppose you have a lot more confidence in the tools than I do. I think we will get bit by this someday, but I have nevertheless done what you want. We now have all separate algorithm packages. However, I decided against Monticello prereqs; one change to Base and you suddenly have to resave *eleven* packages, and we all know how buggy prereqs are. But I'll go any direction the team wants to on this..
I follwed hmm's lead on removing ASN1Constants pool dictionary, it was causing a load warning, but not sure whether it was needed or not. I know nothing about ASN1. Further, I noticed some of the base-class extensions named "decodeAsn1Der" were also used by DSA. Therefore, ALL ASN1 extensions are in the CryptographyBase package. Someone who knows about ASN1 should clean this up.
Remember, you need to load CryptographBase-cmm.3.mcz, NOT version 5. Also, Cryptography-Support is not needed.
After you load Base.3, then load each algorithm-specific package and, finally, the Tests package. 15 of 18 of the tests pass, just as if I load Cryptography 0.3 from SqueakMap.
- Chris _______________________________________________ Cryptography mailing list Cryptography@lists.squeakfoundation.org http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/cryptography
_______________________________________________ Cryptography mailing list Cryptography@lists.squeakfoundation.org http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/cryptography
Hi folks,
Ron Teitelbaum wrote:
Hans-Martin, Also I fixed errors in the test package, so now only the DES primitive error is left. Were you able to work on that, how's it going?
I've started to add my non-primitive implementation but haven't come very far yet due to time constraints. My code worked with 64-bit integers, while the original (primitive) code uses byte arrays, so I have to rewrite quite a bit. I'll get to it next week (hopefully).
Cheers, Hans-Martin
cryptography@lists.squeakfoundation.org
-
Cees De Groot -
Chris Muller -
Hans-Martin Mosner -
Ron Teitelbaum