Here is my method of directory and file organization.
- /srv - I store all my server files in /srv, including WWW and Samba. This is the standard UNIX directory for "Site-specific data which is served by the system."
Then I place all my genealogy related files in a directory call ./Genealogy. The exact path may depend on how you are set up. Here are three different possibilities.
/srv/Genealogy/ /srv/Samba/Genealogy/ - If you use a samba file share. /srv/Samba/Family/Genealogy/ - This has the /Genealogy in a file share of /Family.
This is the structure of ./Genealogy:
Basically there are three main areas. Books or publishing. Documents, which are the Media storage area. Gramps.
- ./Books_Published/ - All books that I have published, such as "Perkins Family History".
- ./Books_In_Process/ - Current book that I am working on.
- ./Doc/ - This is the big one where I store all the Media files. Nothing goes in here till it is put into Gramps Media. See the ./Need_to_Process directory below.
I am big on "Noun Naming". This makes it quick and easy to find documents. Hence for the ./Census directory I use the file name format of <STATE>-<COUNTY>-<CITY><YEAR>-<ENUMERATION>-<PAGE>.jpeg For the Media Title I use Census <COUNTRY>, STATE>, <CITY / TOWNSHIP>, <YEAR ENUMERATION>, p. <PAGE> - <FAMILY>. The Source Tile is the same, except without the , <YEAR ENUMERATION>, p. <PAGE> - <FAMILY>.
This makes it quick and easy to find and cleanup errors. Periodically I do a sort, then quickly look for errors. This includes the path field in the Media.
- ./Doc/Book_<BOOK NAME>/ I have stored documents such as birth certificates, in a three ring binder. These are enclosed in acid free clear sleeves. The "<BOOK NAME>" is the name of the binder. Or you could create a directory for each book.
- ./Doc/Book_<BOOK NAME>/001-<TITLE>.jpeg These are the scans from the book. The first three digits are the page number of the acid free sleeve. Then a title (<TITLE>). This title is the same as I use for the Media title. Hence a path of ./Doc/Book_1/031-Birth_Record_Perkins_Gerald.png has a Media Title of "Birth Record Perkins, Gerald Dana"
- ./Doc/Census/ For a path of ./Doc/Census/california-humboldt-south_fork_township-1920-69-1b.jpeg, the Media title is "Census USA, California, Humboldt, South Fork Township 1920 69 p. 1B", then the Source Title would be "Census USA, California, Humboldt, South Fork Township 1920".
- ./Doc/Head_Shot/ Where I keep an individuals head shot. These are placed in People > Person > Gallery. Hence their picture shows up.
- ./Doc/Media_yyyy/ This is kind of my general catch all. I know, not good. It started to be so large that I started a new one every year. Slowly I have moved some to there proper directories.
- ./Doc/Military/ Example "Registration_Perkins_C_Ray.jpeg"
- ./Doc/Naturalization/ Example "Helwig_Otto.jpeg"
- ./Doc/Object/ These are photos of objects such as "Military_POW_Tag_Helwig_Charles.jpeg"
- ./Doc/Ship_Manifest/ I have been working on the format here. This is a current example sm-philadelphia_rhynland-m-1901-06-16.jpeg. Where philadelphia is the port of entry. And rhynland is the ship. i = Immigration document, m = manifest.
- ./Gramps/ The home of Gramps.
- ./Gramps-yyy-mm-dd/ If I try something that may make a mess of Gramps, I create a backup copy.
- ./Miscellaneous/ Currently store blank Census forms and the like.
- ./Need_To_Process/ Any new files, photos, etc that I need to process into Gramps. Kind of a holding area.
I'm still working this one out but here goes... I have an NTFS partition which I can access from both Windows XP and Ubuntu Linux, that's where I keep all my genealogy files. Because my main computer for doing genealogy work switches between two machines every now and then I name the directory after when I move it to that system. At the moment it's called Database_in_2008-06-20, which I won't type every time, consider './' to mean 'the main directory'.
./Backups/ - has GRAMPS XML files as backups, I do this about once a month and after any large change. ./Dropbox/ - has anything I haven't dealt with yet (it's quite full!) ./Events/$EventNames/ - has media files which are primarily related to an event. ./Individuals/$Surnames/ - has images primarily related to an individual. The files are sorted after the persons (birth) surname. ./Individuals_icons/ - has passport sized images use for generating reports ./Places/$Country/ - has files primarily related to a specific place. I will start subdirectories by country. ./Rapports/$Surname - has some rapports I've sent out. ./Sources/$Source - has all sorts of files which contain direct information, scanned letters and so on... Sorted by the source's source. ./2008-03-11_4.gramps - there are often quite a few of these backups which I go through and put a couple in the ./Backups/ directory and throw the others out.
I don't think this system is so great so I'm working on improving it, see below.
Duncan's planned system
- Maximum number of subdirectories?
- Only base + 7 in ISO 9660 level 1.
- Records covering several family names, like weddings and census records?
- Just the couple's family names.
- Census is of the family, so goes into the dual name directory of the parents, not in an individuals sub directory.
- How should the record properties and directory names be limited so we don't cross the maximum path length restriction?
- Single letter directory names, how do we stop them changing case?
- Indicating unsure dates?
- ca = circa/ approximately, ie: --1810-_-_-ca--
- Is it really necessary with the '-' between date parts? ie: why 1810-12-28 when it could be 18101228?
- One problem is if people forget to add a zero before single digit dates. With the dash we get 1810-01-22 or 1810-01-22 but without it if they forget the zero we get 1810122 which could mean 1810-01-22 or 1810-12-02. I say better safe than sorry.
Problems in progress
- When I want to work on one family with a small portable computer?
- all folders need to under the family name directory, for families under a dual name directory, children under their own family name.
- What about a family portrait not from a known event?
- I think it's best to regard pictures of groups as an event, maybe event=gathering
The base directory is not shown. A name like Imported_2008-12-23 is recommended to record when the data was moved to this machine from some other machine.
Directory naming rules: • Single letters should be lower case, but some windows filesystems will change this without warning. So don't rely on it. • Start words in directory names with capitals, this discourages file systems from changing the case. • Once a family starts (shared address, children or legal union) they join a dual family name directory, alphabetically sorted, ie: Jensen__Williams and not Williams__Jensen.
<first letter>/<family name>/<record type>/ <file> or <given name(s)>/<file>
My Media objects use the same naming structure :
ISO 8601 date_description.extension
./Docs/ numerical sources (scanned certificates, papers, acts, hand written sources) ./Identity/ passport sized images use for generating reports ./Places/ (photos of gravestones, living places or address) ./Other/ not in the first ... (videos, sounds, groups photos)
I will not add subdirectories. Also, all objects are duplicated on an other support (backups and searchs) with my grandparents' surname :
./Surname of the mother of my mother (maiden) ./Surname of the father of my mother ./Surname of the mother of my father (maiden) ./Surname of the father of my father
And I use attributes on my database (as marker). True, using relationships calculator should do the work but it is usefull for filtering and sharing data. Why attributes ? Adding this value (like a description) is not false ... They are and will still be my ancestors on (father or mother side)
-- Romjerome 26 July 2008 (EDT)