[GNC] Writing GCM File

[John Ralls]

David,

WSL works very well and might be more convenient than a separate Linux box.

You might also try editing the gcm file and changing a couple of column widths in one existing account section to see if that sticks.

Regards,
John Ralls

> On Feb 1, 2026, at 07:59, David T. <sunfish62 at yahoo.com> wrote:
> 
> John, 
> 
> Thanks for the suggestions. 
> 
> Knowing how challenging it is to compile on Windows, I'm not sure I'll succeed in following that path, but I'll see what transpires. I do have access to a Fedora box; I think I'll try the same experiment there and see what happens. If it fails the same way, maybe I can compile there and get debug data. 
> 
> I did at one point succeed in compiling 5.13 on that box, so I believe I probably could again. If I get to that point, I'm sure I'll be asking for guidance on setting the necessary compile flag. 
> 
> I did look at the original gcm file and compared EOL characters in my modified version, but I didn't see any differences. Both had Windows CRLF EOL characters. 
> 
> I will note that when I examined the original gcm file, the register information was placed ahead of the metadata for opened tabs (tagged as Page n), and that for my modified gcm, I put all the register settings at the end of the file, to simplify the scripting. I figured that GnuCash would process the entire gcm at startup and the ordering wouldn't matter. I will try modifying the output to restore the original sequence and see whether that changes anything. 
> 
> Thanks, 
> David T.
> 
> 
> On February 1, 2026 1:58:52 AM GMT+05:30, John Ralls <jralls at ceridwen.us> wrote:
>> David,
>> 
>> I don’t see any obvious reason that it would fail as long as you’re sure that you spelled everything correctly. The function that sets the column widths is gnucash_register_configure at https://github.com/Gnucash/gnucash/blob/e0d9d336baee2292b6ddad7ecbc44957baaf4fc6/gnucash/register/register-gnome/gnucash-register.c#L455. GNC_PREF_SAVE_GEOMETRY is set with Preferences>Windows>Window>Save Window Size and Shape. It also controls saving so you must have it enabled or you wouldn’t have found those entries in the first place.
>> 
>> The first thing I’d do would be to try it in the debugger, but it requires full debug symbols and that means building GnuCash yourself with CMAKE_BUILD_TYPE=Debug.
>> 
>> A less onerous approach would be to turn on debug logging for gnc.register. There isn’t a lot of logging in there so it might not help; unfortunately there isn’t any at all in the state file loading code.
>> 
>> Assuming that you’ve carefully compared the two gcm files to ensure that all of the sections and keys are exactly the same, it occurs to me that there may be a line ending difference that would be invisible but might trip up the Glib keyfile loading code and result in the section or keys not being found.
>> 
>> Regards,
>> John Ralls
>> 
>>> On Jan 31, 2026, at 00:41, David T. via gnucash-user <gnucash-user at gnucash.org> wrote:
>>> 
>>> Hello,
>>> 
>>> I am a longtime GnuCash user (19 years) with GC 5.14 (2025-12-10 build) on Windows 11.
>>> 
>>> I was reading a thread recently regarding register column widths, and I decided to try to work out a mechanism to set column widths for every account in my book, by account type. Specifically, I wanted to ensure that any account of type STOCK or MUTUAL had rational column widths for balance, debit, credit, price, and shares.
>>> 
>>> I started by opening an existing gcm file and noting the column width variables. Every register that has had its columns set has a section in the gcm file with values for all columns, identified by the account GUID:
>>> 
>>> [Register XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX]
>>> date_width=133
>>> num_width=44
>>> reconcile_width=33
>>> balance_width=130
>>> transfer_width=600
>>> debit_width=125
>>> credit_width=125
>>> 
>>> Accounts of type STOCK or MUTUAL have additional lines:
>>> shares_width=100
>>> price_width=133
>>> 
>>> I figure that I should be able to write out a gcm file with column settings for every account in the book, and this *should* make it so that any account I open gets my preferred column settings--regardless of whether I have actually set the columns on that register.
>>> 
>>> I created a simple python script that takes a csv list of all accounts in my book (derived from an SQL copy of my data file, and the SQL query "SELECT guid, account_type FROM accounts;") and writes my preferred column settings into a new gcm file.
>>> 
>>> For full disclosure, I started with a copy of the existing gcm file, stripped out all register-related data, and then pushed new register settings combined with this base gcm data into a new text file.
>>> 
>>> In theory now, if I put this new gcm file into the proper location (in my case, c:\users\sunfi\AppData\Roaming\Gnucash\books\myfile.gcm) and fire up GnuCash, I would expect every account to open with the new preferred column settings.
>>> 
>>> Unfortunately, that doesn't seem to happen. Once GnuCash opens, it seems to rewrite every single entry in the gcm file to some preset values based on some sort of unspecified criteria, undoing the changes I inserted. So, for example, if I set debit_width in every case to be 125 (as above), once I open GnuCash, the debit_width *in every single entry* gets reset to 88.
>>> 
>>> Do any devs have insight into what is going on here? Why doesn't GnuCash respect its own gcm file settings? Why rewrite this file in this way? Is it even possible for me to set column widths in the gcm file and have them persist?
>>> 
>>> I would really like to have my register settings a certain way, and going through a thousand registers one by one to reset the columns is not viable.
>>> 
>>> -- 
>>> David T.
>>> 
>>> _______________________________________________
>>> gnucash-user mailing list
>>> gnucash-user at gnucash.org
>>> To update your subscription preferences or to unsubscribe:
>>> https://lists.gnucash.org/mailman/listinfo/gnucash-user
>>> -----
>>> Please remember to CC this list on all your replies.
>>> You can do this by using Reply-To-List or Reply-All.
>>