Global Searching Across Databases Question

Posted by Fredy
Nov 1, 2011 at 05:46 PM

As we all know, I post in the UR forum under schferk, pearls before Schweine and Ferkel (my tongue-in-cheek humour, doesn’t matter), and just some days ago, I’ve been developing this subject there.

The corresponding MI functions are “search all databases” and “search all OPEN databases” ; I had left MI at the beginning of 2011, falsely thinking UR was superior in the end, and indeed, MI’s distinction between current, open and all databases (in a given directory that is) is one of its finest feature, even when “search all databases” takes a lot more time than you’d expect if your data grows exponentially, and considering the fact that MI, as UR, is constantly indexing your data (which slows down its core functioning).

In the UR forum, I got to the conclusion that it’s not “MI or UR” or whatever, but splitting up your data into self-contained files, and do the inter-database search by a third party program, i.e. - far superior to anything else I’ve trialled - Agent Ransack which is free and first-rate : It even gives you the contexts of your finds (which are very readable if you search your ActionOutline files, e.g., weird special chars notwithstanding if you’ve got text with UTF-8 encoding, i.e. “European chars”).

The advantage of such a systems is imo that it’s scaleable in two directions, you can have multiple files on your 700 g tablet comp, and a 10k staffed corp could build up the same system, all with the same ease - try to apply UR (or, worse, MI) to any, just the slightest, corporate environment, and it’ll fail you by their architectural misconception.

Another big advantage of that “keeping it simple” approach to files, together with cross-files search by (excellent) third-party software, is the fact that then it’s possible to build up a GTD (or any other PM) system, because an excellent third-party software’s able to find many special chars you’d put into your files as action codes, whilst in MI and in UR then, I tried several dozens of special chars with their in-built search functions (all 3 with MI, just the 1 in UR), and it had been only a handful of such chars MI had been able to detect, and mostly within special char combinations only (!), whilst UR had been even worse, it only detected 2 (!) special chars out of those several dozen, and in special, unhandy combinations only, in the way (I’m not looking it up now, so this is just a take-place and not necessarily a real-world example): 5$ was searchable, whilst $ was not, nor was $5.

This being said, I’m aware that after you’ll have found your find in any third party program, you’ll need again search it within the given data file, whilst integrated search functions allow for clicking on the result, and you’ll get to the item / find in context immediately then.

But then, neither MI nor UR (within its 1-file search now) will give, up to now, those contexts, whilst AR will give them, which is to say, in MI / UR / whatever, you’ll get some / a lot of FALSE finds, to click then one by one, in order to see the context only then - and whilst in MI you’ll click within a list, and again, in UR those clicks will HIDE your search, and you’ll have to search anew if I understood that function well - it’ll be a lot of clicking back of forth anyway.

So, going down the results list “with context” in AR, finding the really relevant results, and then search again within that given target file - a macro should do the manual work here, having put your original search term not only into AR, but into the clipboard as well -, seems to be the preferable solution of these problems indeed.

As you can see here (and as I could show you with almost any other aspect of those progs), “integrated progs” like MI and UR are just not sophisticated enough to live up to their promises, and thus, combinations of simpler but specialised and faultless softwares could bring better results in more simple ways.

( It’ll be the day AR’s paid brother will find “European chars” I’ll happily buy ; since it doesn’t yet, the freeware does it for me. )