Kaare.Rasmussen@f15.n231.z2.fidonet.org (Kaare Rasmussen) (08/21/90)
Hello All Here's something I picked up on BIX. Hope it's of general interest: I have been reading the previous comments on the Version 2.0 Modula-2 system with increasing distress and thought I had better post a reply direct from the R & D team. Firstly, we admit that the system released has some severe and serious bugs, more than usually expected for the first release of a new, largely re-written product. For this we have no real defence and can only apologize sincerely to those adversly affected. There was a beta-test of more than 50 people worldwide lasting several months, but a number of bugs were introduced (as the result of other enhancements/bug-fixes) at the end of the beta-phase and consequently escaped detection until too late. The product was released, with the benefit of hindsight, much too early. The pressures to release a new version as soon as possible were considerable (not least from impatient customers) and this lead, almost inevitably, to a "rush-job". We hopefully have learnt our lesson and won't make the same mistakes again. Secondly, the quality of documentation in 2.0 is also not up to that expected of a TopSpeed product. There are numerous typos, errors and omissions that did not get picked up by internal proof-reading. Also the explanations of a number of key features are often either misleading or absent. The tight time-schedules mentioned above are partly to blame for this as well as our failure to do extensive external proof-reading and beta-test of the documentation. To correct this in future releases we have set up an all new and more professional documentation team (the person previously responsible for the documentation is no longer with the company). The new manuals will be fully beta-tested and proof-read by professionals before they ship. Thirdly, I would like to comment briefly on the aims and directions of the new TopSpeed compilers. It has previously been noted that multi-language integration is no easy task (and we at JPI can confirm this!), but this is what we are striving to achieve. Not just to allow Modula-2 programmers to use C but also to allow C (and later Pascal) programmers to experiment using Modula-2 with all its attendant benefits. We are still Modula-2 evangelists at heart (although with a certain pragmatic streak (as often pointed out)). Our internal development is still 95% Modula-2 with the rest in TS assembler. The goal of multi-language integration has many consequences. The system is necessarily more complex supporting all of the features of each new language added. To share the same libraries across separate languages requires a common core (which must be written in assembler (note that none of the M2 library is writen in C as has been suggested)) that takes into account the differing library semantics of each language. This must all be done without significantly penalizing users of any particular language. This we have (initially) partially failed as a few C restrictions have adversly affected the Modula-2, but we now hope to have these problems ironed out. The scale of this multi-language integration of libraries (including Pascal and C++ as we are working on compatibility with future products) should not be underestimated. The current complexity of the libraries is a direct result of this, but we are working on a much improved scheme for the next version of TopSpeed which should greatly improve the library structure (and incidentally greatly reduce the number of project files). We are also working on supporting all major C 3rd party libraries although there are a number of problems we are encountering in so doing. ...Continued... Regards Kaare -- uucp: uunet!m2xenix!puddle!2!231!15!Kaare.Rasmussen Internet: Kaare.Rasmussen@f15.n231.z2.fidonet.org