Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

A couple years ago I worked as a programmer in a 1-2M SLOC enterprise Delphi codebase for 6 months (I believe Delphi is basically Object Pascal). It let me do my work. Under the hood there is the same simple machine model as C and you can access it ok, but the ecosystem around it is considerably clunkier. 90's OOP craze is very present. The various container libraries are composed of different strata of types and classes that do the same things but with different philosophies that were popular over time. Mix of 0-based and 1-based indexing in various string and array types is one of the most annoying problems that I can recall.

Syntax-wise, it was weird the we had to do a lot of "PFoo" pointer-to-foo style typedefs. Syntax has a few strange "addressing" models builtin, including out vars etc. Allocation using "SetLength" is weird, a function that takes a dynamic array can realloc it but the changes would not (or only when reallocation was required?) reflect in the caller. I resorted to plain pointers and manual memory management, because everything else was too confusing for me.

To wrap up, it has a lot of fluff built in that doesn't scale because of non-orthogonality of features, and some OOP that is hard to use IMHO because it is too implicit. As a C programmer I am biased but I find that the usable parts are the C-like parts. The syntax is considerably more clunky compared to C, with all the keywords that tire the eyes and fingers after writing a couple hundreds lines of code.

On the plus side, I'm not an ecosystem guy but I'm sure the ecosystem has a lot to offer functionality-wise, including extensive coverage of the Win32 API. And the compiler is really fast, I have good memories of it and the (Embarcadero) IDE. A big reason why it's fast is that the syntax and module system are better designed than C as far as allowing efficient implementation.



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: