STROBL%DBNGMD21.BITNET@wiscvm.wis (Wolfgang Strobl 49+228303223) (08/28/87)
> In article <402@root44.co.uk>, jgh@root.co.uk (Jeremy G Harris) writes: > > Does anybody know what the K&R-listed reserved word 'entry' was for? > > Actually, when I saw it, ti looked like a means for an alternate entry point > into a routine. (Yes, I know this violates block structuring...but it's > [somewhat widely] used on 370-architecture mainframes.) > I would like to mention that multiple entry points are an useful feature for implementing abstract datatypes. One implements each access function to a datatype as an entry to the module which implements that datatype. In this respect the concept may violate "block structuring", but it can be used quite "structured", I think. In short: I would like to have it back ;-). Wolfgang Strobl, STROBL@DBNGMD21.BITNET
gwyn@brl-smoke.ARPA (Doug Gwyn ) (08/29/87)
In article <9052@brl-adm.ARPA> STROBL%DBNGMD21.BITNET@wiscvm.wis (Wolfgang Strobl 49+228303223) writes: >I would like to mention that multiple entry points are an useful feature >for implementing abstract datatypes. One implements each access function >to a datatype as an entry to the module which implements that datatype. Excuse me, but "entry" was not implemented in C, so it's not a matter of "putting it back", nor is an unimplemented (and undefined!) feature of the language by any means useful for what you suggest or for anything else. Anyway, your suggestion is already met by normal C function entries; nothing (except complexity) is gained by multiple function entry points. C++ provides much better "package" facilities than straight C and is worth looking into.
STROBL%DBNGMD21.BITNET@wiscvm.wisc.EDU (08/31/87)
Doug Gwyn <gwyn@brl-smoke.arpa> writes > In article <9052@brl-adm.ARPA> STROBL%DBNGMD21.BITNET@wiscvm.wis (Wolfgang > Strobl 49+228303223) writes: > >I would like to mention that multiple entry points are an useful feature > >for implementing abstract datatypes. One implements each access function > >to a datatype as an entry to the module which implements that datatype. > > Excuse me, but "entry" was not implemented in C, so it's not a matter > of "putting it back", nor is an unimplemented (and undefined!) feature > of the language by any means useful for what you suggest or for anything > else. Anyway, your suggestion is already met by normal C function entries; > nothing (except complexity) is gained by multiple function entry points. > > C++ provides much better "package" facilities than straight C and is > worth looking into. Excuse me, but you omitted my final line In short: I would like to have it back ;-). I am aware of the fact that the entry keyword was never specified and implemented. My preceding sentence was thought as joke with a serious core (to spell it out: if there had been an implementation of multiple entry points into C functions, I could think of reasonable uses for it). The joke was marked as such. I have no access to an implementation of C++ (it is difficult enough to get a working version of C on MVS here), so your suggestion is welcome, but hard to follow. Moreover, whe have a quite good SIMULA compiler here, so for object oriented programming a tool is at hand. SIMULA is a general purpose, object oriented language (the first one, I think) based on Algol, which I can recommend to everyone :-). Wolfgang Strobl, STROBL@DBNGMD21.BITNET