rjbull, thanks for the warm re-welcome!
I never bought dtsearch, but it's decidedly the best of those search tools: It's all about finding things or not, in proprietary file formats, and especially with accented characters like ü and é and ù, here, dtsearch excels whilst Copernic and X1 are very bad (for standard file formats, X1 seems first-rate though). (For Archivarius, I know many people are fond of it; in my special case, it didn't work well, then crashed...) - As said here or elsewhere, the problem with external search tools is, you then have to go back to your "db" / "pim" / text program, etc., and do another, now more specific search, in order to get to the real "hit", in your application (it occurs to me at this moment that some search tools might be able to have you "go" right to that "hit", from a mouseclick in their hit table, when it's standard progs like Word - but forget this for more exotic file formats).
Both Ultra Recall and MyInfo allow for Boolean search, as does IQ and only SOME other pim's: I remember one which had it, but without "no", and no hit table then; another had the hit table, but no Boolean search, and so on - but it's no wonder that many people use UR or MI in spite all any respective problems of each they otherwise cause.
I once stumbled upon DT/TextWorks, and would be willing to pay 1,200 bucks for a prog that really "has it all", but I discarded it then because of their "ask us for a trial (instead of just downloading it) and for a quote (instead of giving the price) - so I never even got to a screenshot of it, let alone a trial. Then, it's a db, which means it's not a tree superposed upon such a db, as UR and MI and IQ and others are, and even the later AS got trees-on-the-fly (by first line, or by field content - very smart thing, the only prob being that with 5-digit record numbers, this regularly took minutes or even crashed (they dumped their forum because it really become much too much negative feedback from almost everybody). As I today said in my KEdit thread, lately it's MI that seems to leave UR trailing, not because MI was so good suddenly, but because there is steady if slow development, whilst UR don't do much upon their roadmap ("not much" being an euphemism for "nothing" here).
In the web, we use Boolean search all the time, in google (and let alone ebay or specialized sites like Dialog you mention), and most people do it even without knowing: In google, they enter two or three search terms, in order to refine their search from start on: a b is a AND be: people do it intuitively, there. (It's for OR that google asks for some knowledge, since that is an (a,b), far from intuitive but some of us know. Not a is -a, etc., and so it's possible to find things.
Whilst in a non-Boolean pim, you CAN'T search for a b, entering a b there would search for "a b", but not for records with a and b, so these desktop pims are mostly really three steps back from what we use, in the web, all time, even without paying attention.
And there's another thing, many such "basic" desktop pim's do not even allow for searching "just in the tree" / "text only" / both, but invariably search everywhere - but then, it's evident that a search "tree only" will perhaps render 5 hits, from which you choose the right one, whilst the same search "everywhere" will get you 200 "hits" in which then you'll have big problems to identify the one you need, without any possibility to refine your search with a second term that also must be within the same record, since in such progs, as said, a b will not work this way - so you are lucky when you remember a second here that also might be in that record you need, but which is only in 120 "hits"... so no discussion here, if a pim only allows for "normal search and then everywhere", it's to be qualified CRAP, whatever it other qualities might be.
As for Notefrog, if I understand this prog well (without ever having trialled for more than just 2 or 3 minutes or so), it relies exclusively upon searching, since there is no tree: at the left, it's the hit table!
Since I used askSam for almost 20 years or so, from early DOS on, and it got its tree-on-the-fly from version 6 only, I know both worlds: search-only (but in the spectacular AS way), and trees, and I must say, I function with trees, holding together related info, and also offering "a dedicated place" for your info, i.e. within a big tree, you remember, more or less (depending also on the good construction of your tree), "wherearound it must be", and I rely very heavily on this feature, i.e. I "search" for my contents by approaching them physically, by opening up headers, then sub-headers: for me, this is an extremely natural way of getting to info.
On the other hand, my memory for real searching often fails me, and even Boolean search doesn't help too much: I remember a search term: hundreds of hits; I suppose another one should also be in those records (but if I'm mistaken here, I'll inadvertently exclude the record I'm searching for!), and even with the combination, I get too many hits, and then I don't really see a third search time that might have been around there - but perhaps not? And then, with all those more-or-less-synonyms no such current program handles equal!
So I must say that with searching, even with good searching, I've got some problems, hence my interest in sophisticated trees. But of course, searching is of the highest interest wherever you have put something OUT of its tree-heading-subheading "way": somewhere else! There, with a hit table and Boolean search, it's 100 times better than with "just normal search and everywhere": I have to spend several minutes on such a search, sometimes, but I find the thing, in the first case; with only basic searches an 100 hits to then be accessed one by one...
But my point here is, even Boolean search isn't good enough, it should include "semantic search", i.d. half-automatic synonym provision. Meaning: You search for dog, and before searching, the program would list up breeds, "puppy", "cute", "ferocious", whatever, in order for you to decide which of these terms should be searched for (and it's even possible to have some of these in different OR groups).
Couple this with an index, and the prog would only present you such breeds (in our example) that really are present somewhere in your texts, and not unnecessarily clutter this first "what to search table" with search terms, taken from a dictionary but which ain't in your text!
Then, this, for several languages, and for combinations of languages. And finally, you could give the program hit numbers when you work within the "what to work" window, meaning your processing the search terms there will give you real-time results how many hits you'd then get.
I know of a list one very early Dos text db which offered some semantic search (but not in the sophisticated I describe here) - askSam was a best-selling program then and "killed" it, by way of most people then buying AS instead. One of the big ironies here: After having got its then really comfortable market position it then held for years, AS was NOT able to implement any semantic search functionality. So today, we're worse off than we were 20 years before, except for visuals: Of course, Windows ( /Mac ) is pleasant to the eye when Dos gets quickly unbearable because visually at least, we get so much "more" today.
Thank you very much, also for the free Dos progs link. I've encountered another such link, with defunct sw, Windows and Dos combined, like early Wordstar versions, 1-2-3 and such, but wasn't really enthousiastic about these. Your link is for defunct progs that are much more special and much more interesting, incl. Inmagic Plus there, citation: "These are not trivial products." - right they are. Will have a good look into this site!
Btw, this semantic search, google does it all the time for you, and even without taking your advice upon them doing so. Hence the interest of having such a system at home, but with you controlling what's found here, and what's discarded from the results. But no, even those specialised tools, incl. dtsearch, don't do semantic search, let alone let you control it. And this, 35 years after the intro of personal computing. /rant
