Article wrote by: Francesco Di Leo
Actually, I tried them all, from the mythical Beagle to Tracker. Beagle is now in effect a dead project. It was quite interesting, able to search many types of files and miscellaneous information, but personally I was not willing to use so many resources for Mono. Patience, they had developed from the outset with the standard libraries perhaps it might be on all computers. I tried other less known programs, Catfish, Pinot and some other that escapes me, cursed memory, up to lead to Tracker a searching program for files (and information) for Gnome. Until version 0.6 it seemed that someone had finally decided to engage seriously in the implementation of a program that allowed me to not always use the memory to remember which file or e-mail contained the information i was searching for. I was promptly denied (and betrayed).
Not only with the later versions it’s slightly regressed, but the bugs are so many in every new release that I do not recommend it for a “serious” use . I reported a bug a while ago on the incorrect indexing of files whose name or content is made up of letters and numbers arranged in various ways. The problem still exists partly, although I have not yet tried the latest versions (and I do not know when I’ll do it because of dependencies). Google Desktop also suffers the same problem. Disheartened by the existence of many searching files programs that promised much but that showed to be just “demo”, i landed at Recoll, with its Spartan interface. But I did not care, I wanted a program that could index all my files without slamming on the reasons for its failure. I did the usual test drive on “road” and I noticed that it also indexes the files on which the others failed. My search was over.
Recoll consists mainly of program/client “recoll” and the program/daemon recollindex. The latter is responsible for indexing files and can be run as a daemon for indexing files in real time. Depending on the distribution you may also find “recollq”, which is the client command line. Once launched, the client Recoll warns you to proceed with its configuration before starting the indexing. Click on “cancel” otherwise it will indexes all your personal folder. In the window that appears enter the folders and subfolders that you want to index by selecting the button “+” (remember to remove “~”with the “-“). Click “OK” and Recoll will start with the indexing of files in these folders. At the end try to find one or more files that contain a particular word or phrase and instantly you’ll get the search result. For future indexing by hand simply select “Update Index” from the “File” menu.
(Figure 1 – example of search)
As you can see in Figure 1 recoll indexes the contents of different file types along with any metadata. It also provides a text-only preview of the content and metadata index as well as the percentage of relevance, a summary of the content and other informations.
In conclusion, if you are looking for a fairly comprehensive program that helps your memory, Recoll deserves to be seriously tested. For me I used it for some time and it fully meets my needs. It also has a comprehensive manual, thing that should not be underestimated.
Notes and observations:
Recoll need different support programs. Under “Help” select “Show missing helpers”and if the answer is “No helpers found missing” then it means that you have already installed all you need on your Linux box.
Recoll also indexes the email and their attachments but you must enter the full address of the folder that stores e-mails (I use the mbox fomat).
You may encounter problems with Microsoft doc file if the content is formed by two or three words. To be exact Recoll fails for the program used for conversion a better filter could solve this small issue.
Ability to use extended attributes (a feature I have personally tested with the author).
The author addresses seriously the few bugs that occur.
The recoll client is based on the Qt library vers. 4 (personally I use it safely with Xfce) while the programs recollq and recollindex are based on a few standard libraries. The filters used may require other libraries (python, etc.)..
Recoll is highly configurable and customizable, though this is not a priority.
Francesco Di Leo