guendel@exunido.uucp (Andreas Guendel) (02/01/89)
In an adaptation of his PhD-thesis John K. Bennett has published the report "Distributed Smalltalk: Inheritance and Reactiveness in Distributed Systems". There he writes: > We believed that too significant a departure from the > protection free Smalltalk environment would be poorly > received by users. As I have thought a lot about integrating flexible mechanisms of access protection into my prototypical distributed extension of ParcPlace-Smalltalk, I would be interested into your opinions. - Do you think that Smalltalk will envetually build a basis for a distributed multi-user commercial application? - Do you think that this user may have different capabilities to access each other's objects? - Do you think that access protection should be integrated into a distributed Smalltalk System or into the applications? - ... Any other related questions. Please mail your answers directly to guendel@unido.uucp or guendel@exunido.irb.informatik.uni-dortmund.de I will post a summary if there is enough interest.
guendel@exunido.uucp (Andreas Guendel) (02/20/89)
The relative small number of responses to my questions on access protection in distributed Smalltalk-Systems has shown that this is not a favorite subject in the Smalltalk-community (therby supporting Bennett's opinion). But I think there has been enough to justify a small summary, mostly based on the opinions of David Keppel, Donald Mead and Mario Wolczko (thanks): > - Do you think that Smalltalk will envetually build a basis for > a distributed multi-user commercial application? Yes, if better features for distribution and access control are provided. Important: availability and standardisation of these features. > - Do you think that this user may have different capabilities > to access each other's objects? Yes, because it is too easy in Smalltalk to shoot oneself in the foot, other feed have to be protected. Yes, because the desired applications demand levels of access protection. > - Do you think that access protection should be integrated into > a distributed Smalltalk System or into the applications? Yes, the only secure way to provide access protection. Yes, extra complexity of distr. applications hidden as mutch as possible. > - ... Any other related questions. Problem: Garbage collection in systems with access protection. Hint: Several other distr. o.o. languages (Emerald, Clouds, Trellis/Owl). Additionally there have been two questions in a number of responses: 1. Literature on (access protection in) distributed Smalltalk-Systems? Distributed Smalltalk: OOPSLA'86: Decouchant -> Integration into the virtual machine OOPSLA'87: Bennett, McCullough -> 2 articles about integration into the image Access Protection in ...: I do not know any published article, but Mario Wolczko has offered a report on this subject to me, please contact him (mario@ux.cs.man.ac.uk) to ask for it. I think he will send a copy to you. I can send a copy of my report "Access Protection in a Distributed Smalltalk- System" to you. It presents a short description of remote message-sending and has its focus on control mechanisms for incomming messages and the subsequent calculations triggered by them. 2. Available distributed Smalltalk-Systems? A really difficult question: Smalltalk-80 is a quite complex system and it is not likely to develop one's own version. My prototypical system is based on the ParcPlace-Systems implementation and therefore I can not pass it to the community (read the report and make your own extensions). If anybody has a public-domain distributed Smalltalk please POST. To get a better discussion, please send all responses directly to the net.