Open jakob-lilliemarck opened 1 year ago
I ran the query on another machine using postgres 14.9 and then the integer oid
integers were different than in the above result for postgres 15.3. I guess that's not very unexpected given there's a major version in between.
A third attempt running it on my server using postgres:
psql (PostgreSQL) 15.4 (Debian 15.4-1.pgdg110+1)
I get the following list, which is also different altough there's no major version change:
oid | cfgname
-------+------------
3748 | simple
13131 | arabic
13133 | armenian
13135 | basque
13137 | catalan
13139 | danish
13141 | dutch
13143 | english
13145 | finnish
13147 | french
13149 | german
13151 | greek
13153 | hindi
13155 | hungarian
13157 | indonesian
13159 | irish
13161 | italian
13163 | lithuanian
13165 | nepali
13167 | norwegian
13169 | portuguese
13171 | romanian
13173 | russian
13175 | serbian
13177 | spanish
13179 | swedish
13181 | tamil
13183 | turkish
13185 | yiddish
That leads me to believe the integer values of the regconfig oid
are not very stable :thinking: and that representing them as string values in rust code is possible will be less sensitive.
While trying to insert a row using a struct with a
TsConfiguration
field, i run into the following error:The issue seems to be the integer value doesn't refer to any regconfig
oid
on my system. I used to following query to check which valid regconfigoid
values exist:That returned:
Using any of those integers directly in an
INSERT
like below works just fine, but if I use the integers defined on theTsConfiguration
in this crate then I run into just the same issue as I get while running the query through rust & diesel.I'm using postgres version:
I'm unsure if these
oid
values are expected to vary across different systems or if they can be expected to be the same, atleast with major versions, I'll try and look into it. Ifoid
integers are not consistent or are expected to change often or , perhaps using a string value for insertingTsConfiguration
would be more stable?