File* argument order, argument types

From: Andres Freund <andres(at)anarazel(dot)de>
To: pgsql-hackers(at)postgresql(dot)org, Thomas Munro <thomas(dot)munro(at)gmail(dot)com>, Peter Eisentraut <peter(at)eisentraut(dot)org>
Cc: Bharath Rupireddy <bharath(dot)rupireddyforpostgres(at)gmail(dot)com>
Subject: File* argument order, argument types
Date: 2023-02-18 00:52:54
Message-ID: 20230218005254.gma6uzp2my5ln7m3@awork3.anarazel.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

While trying to add additional File* functions (FileZero, FileFallocat) I went
back and forth about the argument order between "amount" and "offset".

We have:

extern int FilePrefetch(File file, off_t offset, off_t amount, uint32 wait_event_info);
extern int FileRead(File file, void *buffer, size_t amount, off_t offset, uint32 wait_event_info);
extern int FileWrite(File file, const void *buffer, size_t amount, off_t offset, uint32 wait_event_info);
extern int FileTruncate(File file, off_t offset, uint32 wait_event_info);
extern void FileWriteback(File file, off_t offset, off_t nbytes, uint32 wait_event_info);

and I want to add (for [1])
extern int FileZero(File file, off_t amount, off_t offset, uint32 wait_event_info);
extern int FileFallocate(File file, off_t amount, off_t offset, uint32 wait_event_info);

The differences originate in trying to mirror the underlying function's
signatures:

int posix_fadvise(int fd, off_t offset, off_t len, int advice);
ssize_t pread(int fd, void buf[.count], size_t count, off_t offset);
ssize_t pwrite(int fd, const void buf[.count], size_t count, off_t offset);
int ftruncate(int fd, off_t length);
int posix_fallocate(int fd, off_t offset, off_t len);
int sync_file_range(int fd, off64_t offset, off64_t nbytes, unsigned int flags);

It seems quite confusing to be this inconsistent about argument order and
argument types in the File* functions. For one, the relation to the underlying
posix functions isn't always obvious. For another, we're not actually
mirroring the signatures all that well, our argument and return types don't
actually match.

It'd be easy enough to decide on a set of types for the arguments, that'd be
API (but not necessarily ABI compatible, but we don't care) compatible. But
changing the argument order would commonly lead to silent breakage, which
obviously would be bad. Or maybe it's unlikely enough that there are external
callers?

I don't know what to actually propose. I guess the least bad I can see is to
pick one type & argument order that we document to be the default, with a
caveat placed above the functions not following the argument order.

Order wise, I think we should choose amount, offset. For the return type we
probably should pick ssize_t? I don't know what we should standardize on for
'amount', I'd probably be inclined to go for size_t.

Greetings,

Andres Freund

[1] https://postgr.es/m/20221029025420.eplyow6k7tgu6he3@awork3.anarazel.de

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Andres Freund 2023-02-18 01:06:49 Re: windows CI failing PMSignalState->PMChildFlags[slot] == PM_CHILD_ASSIGNED
Previous Message Amin 2023-02-18 00:36:25 Share variable between psql backends in CustomScan