[GNC-dev] GnuCash 3 on Linux

Derek Atkins derek at ihtfp.com
Tue Mar 19 11:31:58 EDT 2019


Wm,

Wm via gnucash-devel <gnucash-devel at gnucash.org> writes:

>> On Linux at least those environment variables (not the ones which are merely
>> labels) are not created on installation. They can be and then GnuCash should
>> use the locations defined by them (I haven't verified this mainly because I
>> don't have the need to use it). In principle at least you could define
>> GNC_DATA_HOME and GNC_CONFIG_HOME to point to the book location or a
>> subdirectories of the directory containing the book.
>
> Yes, the problem is the movement of significant files wasn't
> forewarned. I really don't know where some of my charity files are
> now!

I'm sorry, but I am confused.  What files, exactly, do you think you
have lost and don't know where they are?  I ask because GnuCash did NOT
move data files!  All it moved was metadata files.  So what exactly do
you mean by "charity files"?  And in what way are they missing?

And for the record, there *was* discussion about these metadata files
moving.  If you like I can go search the archives to find the
conversations?

[snip]
>>> The place was identifiable relative to the book. 3.x changed that.
>>
>> To a place which is also identifiable relative to the book.
>
> Nope.

I am not sure I understand this denial.  The directory changed from 2.6
-> 3.x, but the file name did not.

[snip]
>> I can conceive of a situation where a number of books use a stock standard
>> account heirarchy and one configures a report to depend on that heirarchy
>> but if account guids are incorporated this becomes of limited usefulness wrt
>> the ability to transfer reports.
>
> Yes!  The reports are useless if separated from the book!
>
> The question is this: who thought it was a good idea to separate the
> reports from the book?
>
> I think it was Geert but may be wrong.

You are wrong.

This was "decided" a long long time ago, like DECADES ago.  Indeed, I
feel this almost pre-dates me!  Yes, Geert restructured the report
saving UI code to make it more user friendly, but that only changed the
front-end, not the back end.  The back end is still the same thing it
has been for a very very long time, well before Geert joined the
project.  Blaming him for that is, IMHO, misdirected anger.

>> The report configuration should not contain the information directly and
>> only have pointers to locate the information which is contained only in the
>> book. Is that not the case?
>
> Nope, the reports contain actual internal account references, my
> report won't work for you and vice versa because your accounts and my
> accounts have different internal references.
>
> Do you see why the placement of relevant files further away from the
> data makes less sense or not?
>
> Maybe someone is still on Geert's side?

I am absolutely on Geert's side, because Geert didn't cause this
failure.  Sure, he didn't fix it when we was working on the saved
reports UI, but he didn't make it any worse than it already was before
he fixed the UI.  And that UI fix was before 3.x!

The only change in 2.6 -> 3.x is changing the directory where these
files get stored.

I'm not arguing that the status quo is correct -- I agree with you (and
I know Geert agrees with you) that storing book-specific data in the
saved reports is wrong.  This is one of the many technical debt issues
we have within GnuCash.  But nobody has offered to implement a fix for
it, and most likely it's not a major sore point for many developers.

I do expect this will get fixed eventually.  I'm hoping in my lifetime ;)

Thanks,

-derek

-- 
       Derek Atkins                 617-623-3745
       derek at ihtfp.com             www.ihtfp.com
       Computer and Internet Security Consultant


More information about the gnucash-devel mailing list