Re: Re: pg_dump new feature: exporting functions only. Bad or good idea ?

Hello David,

thank you for your interest in that patch.

Le ven. 25 mars 2022 à 01:17, David G. Johnston <david(dot)g(dot)johnston(at)gmail(dot)com>
a écrit :

> On Thu, Mar 24, 2022 at 4:42 PM Chapman Flack <chap(at)anastigmatix(dot)net>
> wrote:
>
>> On 03/27/21 08:57, Andrew Dunstan wrote:
>> > We can bikeshed the name of the flag at some stage. --procedures-only
>> > might also make sense
>>
>> Any takers for --routines-only ?
>>
>> "Routine" is the genuine, ISO SQL umbrella term for a function or
>> procedure, and we have used it that way in our docs and glossary.
>>
>>
> Regardless of the prefix name choice neither blobs, tables, nor schemas
> use the "-only" suffix so I don't see that this should. I have no issue if
> we add three options for this: --routine/--procedure/--function (these are
> singular because --table and --schema are singular)
>

Actually, I thought of it after the --schema-only flag (which is kind of
confusing, because it won't export only schema creation DDL). I agree to
keep the flag name singular if we're using a pattern to cherry-pick the
objects we want. My problem is how do you think we could get all the stored
procedures/functions at once? --function=* ? It seems to me that exporting
everything at once is the main use case (I'd be happy to be proven wrong),
and it does not feel intuitive to use --function=*.

How would you see to create 3 flags: --functions-only /
--function=<pattern> / --exclude-function=<pattern> ?
And then we could create the same series of 3 flags for other objects.
Would that be too verbose?

>
> --blobs and --no-blobs are special so let us just build off of the API
> already implemented for --table/--exclude-table
>

> No short option is required, and honestly I don't think it is worthwhile
> to take up short options for this, acknowledging that we are leaving -t/-T
> (and -n/-N) in place for legacy support.
>

I agree

>
> --blobs reacts to these additional object types in the same manner that it
> reacts to --table. As soon as any of these object type inclusion options
> is specified nothing except the options that are specified will be output.
> Both data and schema, though, for most object types, data is not relevant.
> If schema is not output then options that control schema content objects
> only are ignored.
>
> The --exclude-* options behave in the same way as defined for -t/-T,
> specifically the note in -T about when both are present.
>
> As with tables, the affirmative version of these overrides any --schema
> (-n/-N) specification provided. But the --exclude-* versions of these do
> omit the named objects from the dump should they have been selected by
> --schema.
>

That's fine with me.

Have a nice day,

Lætitia

>
> David J.
>
>