Thanks for merging #2264 and #2266 John.

Arrg.  As I've started to poke more into this read-only and lock
topic, I've found some surprises and ultimately a correctness matter.

What SESSION_READ_ONLY actually means
------------------------------------------------------------

1. At the engine level, SESSION_READ_ONLY doesn't actually enforce
read-only. The mode only affects lock handling -- it does not mark the
book read-only. Book write-protection is applied separately by the GUI
only, which calls qof_book_mark_readonly() after opening. I confirmed
with an API bindings test that a SESSION_READ_ONLY session can still
edit an account and have the change persisted to the SQL store. So an
API/bindings consumer that opens SESSION_READ_ONLY expecting the book
to be unwritable does not get that guarantee unless it also calls
qof_book_mark_readonly() itself. [a bug if one interprets
SESSION_READ_ONLY reading the enum name while not carefully parsing
the header comments ]

2.  SESSION_READ_ONLY ignores the lock. A read-only consumer opens a
book even while another process holds it open, perhaps mid-write,
which is exactly when the state can be inconsistent; and because the
SQL backend commits incrementally, a reader can catch a
partially-applied edit. [ a read-only client correctness bug ]

So today the mode neither protects read consistency (it ignores the
lock) nor guarantees read-only-ness (that's a GUI add-on), even though
the name implies both.

Proposed path forward
-----------------------------------

For building read-only tools on the API, what I'd want is a read-only
session mode that (a) takes the lock, so a writer cannot be active
while the tool reads, and (b) actually enforces no writes.

A possible direction would be to replace SESSION_READ_ONLY with two
modes whose names state what they do, and have both mark the book
read-only so enforcement is intrinsic rather than GUI-only:
- SESSION_READ_ONLY_EXCLUSIVE -- takes the lock (fails if the book is
already locked); for tools that need a consistent, uninterrupted read.
Calls qof_book_mark_readonly(). Obviously a SQL backend user would
need sufficient permissions to take the lock.  I'd want to add some
user guide documentation about this where none exists today.
- SESSION_READ_ONLY_IGNORE_LOCK -- opens regardless of the lock state;
the path the GUI's "open read-only anyway" override would use. (i.e.
today's SESSION_READ_ONLY behavior, but with qof_book_mark_readonly()
)

These could be added now, with SESSION_READ_ONLY marked deprecated,
and SESSION_READ_ONLY removed in a later release (v6? whatever that
timeframe turns out to be).  I believe the deprecation warning would
only apply to users of the API `SESSION_READ_ONLY` mode today. (The
GUI internal use would be swapped to SESSION_READ_ONLY_IGNORE_LOCK.)"

Does folding read-only enforcement into the mode (so bindings
consumers get it too), plus splitting the lock behavior, sound like an
OK path?

Aside,  is it a known issue Doxygen isn't updating / is stale? (title
says "GnuCash  5.6-150-g038405b370+" )
https://code.gnucash.org/docs/STABLE/qofsession_8h.html
I was initially looking for the enum descriptions but didn't find them
outside of the code.  Double btw:  There is a documentation bug in the
source here in that the five /** … */ comments need to be trailing
/**< … */ comments (Doxygen trailing-member syntax).  Right now, when
the comments are rendered, they are offset by one.

Regards,

Noah


On Sat, Jul 11, 2026 at 8:47 AM Noah <> wrote:
>
>
> Thanks, John.
>
> > Your premise that the database was necessarily written by a GnuCash session 
> > is false.
>
> You're right, I was biasing by my own use principles.
> I never want to modify my production book through anything but the GnuCash 
> GUI or the official GnuCash API / bindings.
>
> My household moved our finances to GnuCash 14 years ago motivated by two 
> challenges.
> - need to support two users on different computers (in serial, not 
> simultaneously). We were bit too many times with filesystem-based commercial 
> options at the time.
> - a Q* commercial software was dropping support for Mac, or non-cloud, or both
>
> In more recent years, our financial institutions have been making it harder 
> to get our transaction data into GnuCash (eg. removing OFX/QFX download)
> In order to hold the line at home, I've been doing more and more work 
> adopting use of the  API & Python bindings to help a lot in this regard.  
> I've also been creating some reporting and exploration tools that don't need 
> write access (and tinkering with other's work like piecash). Since they don’t 
> need write access, I want them to access MySQL with the minimum privileges 
> necessary.
>
>
> > The ffastmath problem underlying bug 611936 was on the write side of libdii 
> > so a read-only test wouldn’t find the problem.
>
> That settles it for me: since the -ffast-math defect in #611936 is on the 
> write side of libdbi, a read-only session can neither trigger nor detect it, 
> so running the probe on a read-only open guards nothing. I've opened a PR 
> that skips dbi_library_test() for SESSION_READ_ONLY and leaves all 
> write-capable modes untouched:
> https://github.com/Gnucash/gnucash/pull/2266
>
> I'm not trying to bolt access control onto GnuCash, and I agree that wouldn't 
> end well. The narrower goal is just to let a read-only consumer honor a 
> least-privilege credential a DBA may already have in place -- and, 
> independently, to stop running a write-only probe on a read-only open. The PR 
> stands on that second part alone.
>
> Thanks again,
> Noah
>
>
>
> On Thu, Jul 9, 2026 at 8:31 PM John Ralls 
> <[email protected]> wrote:
> >
> >
> >> On Jul 9, 2026, at 14:34, [email protected] wrote:
> >> Hello again,
> >> Obviously backend stuff is on my mind lately and there's been a
> >> quirk impacting read-only access to SQL back-ed books that has
> >> bothered me for a while.
> >> Opening a book with SESSION_READ_ONLY through the SQL/DBI backend
> >> still requires the database user to have permission to create a
> >> temporary table. On every session_begin (regardless of mode),
> >> dbi_library_test() runs the 64-bit round-trip guard from Bug #611936
> >> -- CREATE TEMPORARY TABLE / INSERT / SELECT / DROP -- to detect a
> >> 15-year-old -ffast-math libdbi miscompilation. A least-privileged,
> >> SELECT-only database user therefore cannot open the book at all: the
> >> create fails and begin returns ERR_SQL_DBI_UNTESTABLE.
> >> (SESSION_READ_ONLY already skips the gnclock lock, so this probe is
> >> the only remaining reason a read-only open needs write-ish rights.)
> >> I'd like to propose making a read-only open work without those
> >> privileges. Two possible directions:
> >> 1. Skip dbi_library_test() when the session is opened SESSION_READ_ONLY.
> >> 2. Or, for read-only, replace it with a check that needs no write
> >> privileges -- for example, verify the driver retrieves large
> >> int64/uint64 and a high-precision FLOAT8 via a SELECT of
> >> literals/casts (dbi_result_get_longlong / get_ulonglong / get_double),
> >> so the reader's conversion path is still validated. I'd welcome
> >> suggestions on the best minimal form.
> >> The justification for either approach: the data already in the
> >> database was written by a GnuCash session, and dbi_library_test() ran
> >> for that session, so any book that was successfully saved was
> >> necessarily saved by a driver that passed the round-trip test -- the
> >> stored 64-bit and FLOAT8 values are therefore correct at rest. A
> >> read-only session never writes back, so the only thing left to guard
> >> against is a reader-side retrieval defect. For a modern 64-bit libdbi
> >> that is the 2011, 32-bit / -ffast-math case from #611936, which is
> >> essentially historical (or maybe never existed on the read)
> >> If there is a preferred direction, I'm happy to prepare a PR.
> >
> > Your premise that the database was necessarily written by a GnuCash session 
> > is false. Anybody with db write access (and for a SQLite3 database that 
> > means anyone with write access to the filesystem) can write (or delete) 
> > whatever they want into the database. There is at least one public Python 
> > library that reads and writes GnuCash SQLite3 files 
> > (https://github.com/sdementen/piecash). The databases have no integrity 
> > checks enabled: Any integrity checking (and it’s pretty minimal) is in 
> > GnuCash’s engine.
> >
> > The ffastmath problem underlying bug 611936 was on the write side of libdii 
> > so a read-only test wouldn’t find the problem. I think it unlikely that 
> > anyone is using that version of gcc anymore so it’s probably safe to remove 
> > the test.
> >
> > That said, I have my doubts about the value of having users without 
> > database write access. Access control hasn’t ever been a consideration in 
> > GnuCash’s design and trying to bolt it on 30 years later isn’t likely to 
> > produce a satisfactory result.
> >
> > Regards,
> > John Ralls
_______________________________________________
gnucash-devel mailing list
[email protected]
https://lists.gnucash.org/mailman/listinfo/gnucash-devel

Reply via email to