On Sat, 2014-01-11 at 21:22 -0500, Adam Jensen wrote: > On Sun, 12 Jan 2014 01:07:58 +0100 > [email protected] wrote: > > > Le 2014-01-11 16:26, Adam Jensen a écrit : > > Some years ago, I have written an article about the genesis of GHDL > > and interviewed Tristan. It was published in french in Linux Magazine > > France, > > I forgot the details. But Tristan is definitely up to the image that > > his source code conveys :-) > > > > I would like to know more. A quick calculation: > > 10 lines of code / 8 hour work day > 250 work days / year > > (135k LOC) / ((10 LOC / day) * (250 days / year) = ~54 years
I'm going to say that in my opinion, - without diminishing Tristan's achievement - it owes a lot to his choice of Ada. I find that even for tiny hundred-line hacks, Ada gets me a lot closer to a working solution a lot faster than (insert popular lang choice). because the language really "does what it says on the tin". Browse the bug reports for one aspect if you're curious : https://gna.org/bugs/?group=ghdl (and turn on "all" instead of "open") and compare with any other Open Source project. In the compiler I can't recall seeing a single buffer overrun or access (pointer) violation, and one lone integer overflow (which instantly raised an exception instead of corrupting anything else). You might find a couple of these in the generated code, which is a different problem... What's left is mostly : the user submitted invalid VHDL and the compiler aborted cleanly, but with a stack dump instead of a useful error message. Write fewer bugs and you spend more time programming, less time debugging... That and Tristan's rigorous approach to quality : checking object types everywhere and raising exceptions as soon as anything unknown appears, catches misunderstandings early, allowing them to be corrected ASAP. The choice of language is (IMO) just one aspect of this. And as compilers go, it's still complex but it's a lot more readable than, say, other parts of gcc... - Brian _______________________________________________ Ghdl-discuss mailing list [email protected] https://mail.gna.org/listinfo/ghdl-discuss
