Man good, Info bad   by Eugene Reimer, 2007-Oct04

Whenever I see a man-page containing words to the effect that "This man-page is out of date and incomplete; for correct and complete information see the info-pages", I find myself asking questions like "why are we here?"  or "how did get to this sorry state of affairs?"  or "what was Richard Stallman thinking?"  or "why hasn't anyone written an info-to-man converter?" 

Long before there was info, when the Computer-Science department at our university acquired a PDP-11 running Unix, they made printed copies of the entire set of man-pages; one such set being a stack of 8.5x11 pages about one-half inch thick, that sold for 50-cents to cover the printing costs (it came stapled but you had to unperf the pages yourself).  Before I get all misty-eyed with nostalgia for the good-old-days when Unix was small, let me get back to the point.  The point is that man-pages were fine back when Unix was small, man-pages are still fine now that GNU-Linux is big, which means that info is a solution to a non-problem, a fix to that what aint broke. 

I think we've all come to appreciate that when it comes to finding things on the world-wide-web then the unstructured approach of the word-indexing search-engines is more helpful than any hierarchical index no matter how carefully constructed it might be.  This observation would, I suspect, have seemed profound to most people fifty years ago; it doesn't today due the widespread use of tools such as Google.  But it does still illustrate that information cannot be forced into a hierarchy without suffering a good deal, indeed methinks it ought not to be so forced.  And that is why info is bad. 

Consider how easy it is to find the things you want in a large man-page such as the one for bash.  Then try to find something in an equivalent tree of info-pages.  Doing a thorough job of such a comparison would mean a substantial investment of time, and I'm afraid I gave up on info before becoming altogether proficient at using it.  I did however give the matter some thought before reaching my conclusion. 

Much the same comparison can be made with web-pages.  Consider a situation where you have a choice between getting a document as one large page, or as a table-of-contents having clickable links to smaller pages.  Web-site builders frequently lack moderation, and one taste of the TOC sort of hierarchy seems to lead to more and more levels of it, until finally the actual information is packaged into tiny little morsels, and the reader is forced to spend most of his/her time struggling with navigation, frequently getting lost amongst all those up-, down-, sideways-steps.  This leads to much frustration and swearing but little that is productive. 

By the way, I think Richard Stallman a brilliant fellow who has done a lot of good for free and open software.  Also a stubborn fellow.  While that stubbornness has frequently served him in good stead, it also makes it hard for him to see that one of his ideas was a bad one.  What the GNU-Linux world needs is for Texinfo (makeinfo) to offer Manpage as one of its output formats.  And for the info viewer to be discarded. 

PS:  Another dubious thing Stallman has given to the world:  the rules about how the GPL-prose must be included in its entirety in every file to be so licensed.   Being forever confronted with that mind-numbing prose is just so precisely the exact opposite of what a programmer would want, that I wonder why he made such a rule.  Because he knew it was that important.  He was wrong;  property-rights are important but not that important.  Furthermore, the human mind has an amazing ability to tune out the oft-repeated, which makes him wrong on a different level too.  

For most of us, the too often repeated becomes something we very nearly ignore -- not altogether mind you, it does still distract us from the work we're trying to do, it does some damage, mostly by preventing us from seeing that about each program that really is vitally important (which is how this program differs from all others), and to the extent that this blocking out is incomplete, we are annoyed, which means that our natural tendency is to think ill of the GPL.  Not at all what Stallman had in mind.