looking for Linux software to check my Gedcom

[mariannevanharten]

Recent I switched from Windows to Linux (Ubuntu).
I am looking for a replacement of GedcomValidator to check my Gedcom files.

Who is able to help me?

Best regards,
Marianne

[dthaler]

Linux support for GedValidate is in progress.
(There's an Ubuntu file at https://github.com/ArmidaleSoftware/gedcom7/releases/tag/v1.6.1 but I don't think it's correct yet... it will be .tar.gz but I think the in-progress step has a .zip with both windows and linux binaries in it, so stay tuned)

[tychonievich]

I run Linux exclusively. Of the validators we have listed on https://gedcom.io/tools/#development-aids, I had success running the Python command-line tool https://github.com/cartwrightdj/gedcomtools and in downloading https://github.com/gedcom7code/js-gedcom/ and opening its index.html as a file:// URL in my browser to use it too. I've not tried the others there locally.

[dthaler]

GedValidate v1.6.2 should now run on Linux. Let me know if any problems are experienced.

[mariannevanharten]

@dthaler I uploaded my Gedcom file (51.5 MB). This file is too large...
Also paste the contents isn't possible.

[dthaler]

Thanks for the report. If you can send me a link to the file, or any other similar file if you're worried about privacy, to test I'd be happy to reproduce and try to fix. My email address is in my github profile.

[mariannevanharten]

Hi Andrei,

First of all:
When I send a reply using my mail client (Thunderbird) it will be sent to notifications@github.com. That mail address replies with:
The recipient's mailbox is full and can't accept messages now. Please try resending your message later, or contact the recipient directly.

Because I don't know your mail address, I will reply on Github. I wrote:
Hello Andrei,

Thanks for your mail.

Just now I was busy my Gedcom file, so I could test your app immediately.

The first thing I noticed: much more "errors" than GedcomValidator reported...

A snippet of my gedcom file:

1 EVEN
2 TYPE Transport
2 ASSO @i23@
3 ROLE OTHER
4 PHRASE notaris
2 NOTE obligatie
2 SOUR https://www.openarchieven.nl/saa:5833c92c-9ee6-e29b-6750-a9badeecdc79

reported 2 errors: "Missed value for EVEN" and "Value for SOUR should be POINTER". I don't understand why I see this messages.

Furthermore: nice app!!

Vriendelijke groet,
best regards,

Marianne van Harten

[lavich]

Hi Marianne,

I wrote a GEDCOM editor called Domorium, and the size problem you hit is the one
thing it was recently built for, so it may be worth a try:

https://domorium.com/ — nothing to install, works on Ubuntu in any modern
browser. Use Open to pick your .ged. It runs entirely in the page: the file
is read locally and never uploaded anywhere.

I tested a generated GEDCOM 7 file of the same size as yours — 51.5 MB,
2.7 million lines, 192k individuals and 43k families all cross-referenced. It
opens and shows its first diagnostics in about 3 seconds, and the document is
held whole (saving a copy back gives a byte-identical size). Both GEDCOM 5.5.1
and 7.0 are supported.

Three honest caveats, because a file that size is near a browser's limits:

• It wanted about 1.4 GB of memory for that document. On a machine with
  little RAM the tab may not manage it.
• It shows problems in the editor, on the line that caused them, rather than
  as a printable report. If a report is what you liked about GedcomValidator,
  this is a different thing.
• It does not yet catch a day the calendar does not have — 31 FEB 1880 passes
  today. I've filed that; the malformed-date shapes it does catch are things like
  FEB 32 1880 and 1880-01-01.

My test file was generated, so it is more uniform than a real export. If yours
behaves differently, or does not open at all, I would genuinely like to know —
an issue at https://github.com/lavich/domorium/issues with what you saw is
enough, no file needed.

GedValidate is worth staying with too if you prefer a desktop tool; it sounds
like the size limit there is being looked at.

(Domorium is an independent project, not affiliated with FamilySearch.)

[lavich]

Hi Marianne,

Thank you — you found a real bug within hours, which is more than my own test
files managed.

Missing value for EVEN is my mistake. The specification says a payload may
be omitted whenever its data type allows the empty string, and EVEN carries a
Text payload, whose grammar is *anychar — zero or more characters. So
1 EVEN with only a TYPE beneath it is perfectly valid and I should not have
flagged it.

It is not just EVEN: the same mistake affects 61 of the 182 payload types, so a
good share of the extra errors you saw are mine rather than your file's. AGE and
DATE are wrong in the same way — the specification explicitly allows leaving
them empty and giving a PHRASE instead. I've filed it and will fix it shortly.

Value for SOUR should be POINTER is, I'm afraid, real. In GEDCOM 7 a source
citation has to point at a source record rather than carry text:

0 @S1@ SOUR
1 TITL Open Archieven
2 SOUR @S1@
3 PAGE https://www.openarchieven.nl/saa:5833c92c-9ee6-e29b-6750-a9badeecdc79

The rule is at
SOURCE_CITATION in the specification, and the structure definition at
https://gedcom.io/terms/v7/SOUR states it outright:

payload: "@<https://gedcom.io/terms/v7/record-SOUR>@"

The @…@ means a pointer to a record, so a URL in that position is not valid
GEDCOM 7. In a conforming file the link would sit in a PAGE beneath the
citation, as above, or in a WWW on the source record itself.

One thing worth knowing if you go looking, because it confused me for a moment
too: there are three different structures spelled SOUR, and which one applies
depends on where it sits. 1 SOUR in the header identifies the program that wrote
the file and takes text — a URL there is fine, and expected. Only the SOUR
beneath an event or record is a citation and needs the pointer.

I'll also improve the wording — "should be POINTER" describes the rule rather than
the problem, and you are the second person to find it unclear.

I'll reply here once the payload fix is out, so you can re-check the file and see
what is left. Thank you again for taking the time.

[lavich]

@mariannevanharten

The payload fix is live — nothing to install, just reload
https://domorium.com/ and open your file again.

1 EVEN with only a TYPE beneath it is accepted now, along with the other 60
payload types that had the same mistake, and AGE or DATE left empty with a
PHRASE instead. The SOUR message is reworded too — it says what belongs in
the slot rather than naming the rule.

Two things changed in dates, and they pull in opposite directions.

I had been rejecting every GEDCOM 7 date that names its calendar — JULIAN 1401,
HEBREW 1 TSH 5761, even GREGORIAN 1 JAN 2000. That is fixed, so some errors
will simply disappear. I found it by running this repository's own test files:
testfiles/gedcom70/date.ged has 1062 DATE lines and I was rejecting 942 of
them.

The same work made me stricter about three forms 7.0 removed, though:
@#DGREGORIAN@ 9 MAR 2007, a slashed year like 1 JAN 1857/58, and
INT 1950 (around 1950). If your program writes any of those, you will see date
errors that were not there yesterday. As far as I can tell they are correct — but
if you get thousands of them, please say so and I will check whether it is your
file or my reading of the specification.

Thank you again for the report.