Hi Daniel,
I agree with you if psql output doesn't indicate any error when the disk is
full, then it is obviously not nice. In some situations, people may end up lost
data permanently.
However, after I quickly applied your idea/patch to "commit
bf65f3c8871bcc95a3b4d5bcb5409d3df05c8273 (HEAD -> REL_12_STABLE,
origin/REL_12_STABLE)", and I found the behaviours/results are different.
Here is the steps and output,
$ sudo mkdir -p /mnt/ramdisk
$ sudo mount -t tmpfs -o rw,size=1M tmpfs /mnt/ramdisk
Test-1: delete the "file", and run psql command from a terminal directly,
$ rm /mnt/ramdisk/file
$ psql -d postgres -At -c "select repeat('111', 1000000)" > /mnt/ramdisk/file
Error printing tuples
then dump the file,
$ rm /mnt/ramdisk/file
$ hexdump -C /mnt/ramdisk/file
00000000 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 31 |1111111111111111|
*
00100000
Test-2: delete the "file", run the command within psql console,
$ rm /mnt/ramdisk/file
$ psql -d postgres
psql (12.1)
Type "help" for help.
postgres=# select repeat('111', 1000000) \g /mnt/ramdisk/file
Error printing tuples
postgres=#
Then dump the file again,
$ hexdump -C /mnt/ramdisk/file
00000000 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 | |
*
00100000
As you can see the content are different after applied the patch.
David
The new status of this patch is: Waiting on Author
