Front page | perl.perl5.porters |
Postings from December 2013
Posting to P5P, because this pertains not only the article, but XS itself: As i went on reading the article, and playing with the code in the github repo, more and more questions about the cryptic names and bizarre idiosyncrasies of the code came to me. Eventually i realized and understood that they were no specific choice in implementation, but forced due to the cryptic, bizarre and idiosyncratic naming and concepts of XS itself. * Discussion with Zefram in #p5p failed to show me any other possible reason, since he himself tentatively agreed that the arguments boil down to "tradition". The trouble with XS macros is that they generally carry names that do not actively describe the origin, content or purpose of variables; and to make matters worse, have names that are nearly useless as search queries. They thus require the dedicated seeker to try and acquire the knowledge of which specific box to look in for the explanation of those macros and subsequently memorize all of them, simply in order to acquire the ability of reading the code. ** This is the exact same trouble that ails magic variables, and could be similarly addressed with an appropiately crafted `include 'xs_english.h';`. I'm willing to do this experiment, however my experience and exposure to XS is limited, so i am thinking i will just do it in the repo of the code to this article. -- With regards, Christian Walde * A very simple and hopefully extremely clear example is the variable `items`. It is, exceptionally, an english word in XS, whose meaning can be guessed. However even so, that's only a guess. A simple renaming to sub_arg_count would make it entirely unnecessary to guess. ** And as we all know, code should be written such that the next person damned to maintaining it can read it with least effort.Thread Previous | Thread Next