(d) It makes it easier to work with a fixed-width terminal or terminal emulator.
You may think this is an obsolete requirment, but I spend most of my working days using a VT-100 emulator connected to various servers running just Oracle, gdb, gcc, vi, and a shell. At 09:34 AM 3/6/2003, Paul Beardsley wrote: >Hello, > >I have a question about coding style which maybe fits the list criteria of "technical >discussion about a library'. > >The Boost style guide recommends 80 column text without giving any justification for >this. What is your opinion of this constraint? > >The three supporting arguments I hear are > >(a) It makes it possible to print code. Personally I never print code. I understand >it might be useful given, say, 15 people in a room doing a code review but even then >I think it would be better to bring in laptops and PCs and look at the electronic >version. For one thing at least then everyone ends up with the same annotations. > >(b) It makes it possible to do side-by-side comparisons of two slightly differing >pieces of code. Occasionally I need to do a comparison but I really try to avoid it >- there is something wrong if the development process keeps throwing up the need for >a laborious (error-prone) manual comparison of pieces of code. > >(c) Reading test over 80 columns makes your eyes tired, that's why typesetters >traditionally converged on the 80 column solution. I think that reading >highly-structured code is different to reading straight text, so I don't know if the >lessons from straight text carry over. > >My arguments for longer-than-80-column lines would be the following. These arguments >apply specifically to header files - > >(i) It's easier to parse a header file with one method per line. Longer lines make >this more likely. > >(ii) It's preferable to have inline method definitions with the declaration, for >short methods. Longer lines again make this more likely. > >(iii) The structure of the code throws up other cases where one method per line makes >for readability e.g. overloaded methods benefit from being on consecutive lines so >that the first part of the text corresponds directly and only the signatures differ. > >The most powerful argument in favor of 80-column text is just that 'everyone does >it', or that you cannot anticipate that somebody-somewhere might have a device that >needs 80-column text. But these seem more like arguments by inertia and surely >readability is more important. So I wondered what the general opinion of boost >developers is on this issue, > >Thanks, > >Paul Beardsley. > > > > > >Do you Yahoo!? ><http://rd.yahoo.com/finance/mailtagline/*http://taxes.yahoo.com/>Yahoo! Tax Center - >forms, calculators, tips, and more _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost