Thread: per feedback I received directly, pg_dump manpage patch

per feedback I received directly, pg_dump manpage patch

From
"Joshua D. Drake"
Date:
The man page does not clearly state that pg_dump will not dump things
like users or roles. It does mention briefly in the notes section that
pg_dumpall is needed for items that are set via SET but that isn't very
clear on the other items.
--
Command Prompt, Inc. - http://www.commandprompt.com/  503-667-4564
PostgreSQL Centered full stack support, consulting and development.
Announcing "I'm offended" is basically telling the world you can't
control your own emotions, so everyone else should do it for you.

Attachment

Re: per feedback I received directly, pg_dump manpage patch

From
Alvaro Herrera
Date:
Joshua D. Drake wrote:
> The man page does not clearly state that pg_dump will not dump things like
> users or roles. It does mention briefly in the notes section that pg_dumpall
> is needed for items that are set via SET but that isn't very clear on the
> other items.

I agree that we need a warning about this, but I think it should be a
bit more generic rather than focused exclusively on pg_dumpall; perhaps
something like "global objs such as roles/tblspcs should be backed up
separately, for example using pg_dumpall".

I didn't check the location of the warning in your patch, or whether a
warning is really appropriate rather than, say, <note> or a plain
<para>.

--
Álvaro Herrera                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services


Re: per feedback I received directly, pg_dump manpage patch

From
"Joshua D. Drake"
Date:
On 05/27/2015 06:31 PM, Alvaro Herrera wrote:
>
> Joshua D. Drake wrote:
>> The man page does not clearly state that pg_dump will not dump things like
>> users or roles. It does mention briefly in the notes section that pg_dumpall
>> is needed for items that are set via SET but that isn't very clear on the
>> other items.
>
> I agree that we need a warning about this, but I think it should be a
> bit more generic rather than focused exclusively on pg_dumpall; perhaps
> something like "global objs such as roles/tblspcs should be backed up
> separately, for example using pg_dumpall".

As the only way to back up global objects is pg_dumpall....?


>
> I didn't check the location of the warning in your patch, or whether a
> warning is really appropriate rather than, say, <note> or a plain
> <para>.

The warning is directly after the first paragraph. I thought about using
<note> but I think this does warrant a <warning>. You can't actually
restore a database correctly without pg_dump + pg_dumpall (in this context).

JD



--
Command Prompt, Inc. - http://www.commandprompt.com/  503-667-4564
PostgreSQL Centered full stack support, consulting and development.
Announcing "I'm offended" is basically telling the world you can't
control your own emotions, so everyone else should do it for you.


Re: per feedback I received directly, pg_dump manpage patch

From
Alvaro Herrera
Date:
Joshua D. Drake wrote:
>
> On 05/27/2015 06:31 PM, Alvaro Herrera wrote:
> >
> >Joshua D. Drake wrote:
> >>The man page does not clearly state that pg_dump will not dump things like
> >>users or roles. It does mention briefly in the notes section that pg_dumpall
> >>is needed for items that are set via SET but that isn't very clear on the
> >>other items.
> >
> >I agree that we need a warning about this, but I think it should be a
> >bit more generic rather than focused exclusively on pg_dumpall; perhaps
> >something like "global objs such as roles/tblspcs should be backed up
> >separately, for example using pg_dumpall".
>
> As the only way to back up global objects is pg_dumpall....?

Well, my point is that the roles could be set up separately by other
means.

> >I didn't check the location of the warning in your patch, or whether a
> >warning is really appropriate rather than, say, <note> or a plain
> ><para>.
>
> The warning is directly after the first paragraph. I thought about using
> <note> but I think this does warrant a <warning>. You can't actually restore
> a database correctly without pg_dump + pg_dumpall (in this context).

Okay, +0 there.  (That is, I don't disagree with you but I won't stand
in the way of others disagreeing on that point.)

--
Álvaro Herrera                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services


Re: per feedback I received directly, pg_dump manpage patch

From
"Joshua D. Drake"
Date:
On 05/27/2015 06:53 PM, Alvaro Herrera wrote:

>> As the only way to back up global objects is pg_dumpall....?
>
> Well, my point is that the roles could be set up separately by other
> means.

True but this is about pg_dump right? I mean contextually it doesn't
matter if they can issue a separate SQL command as we are talking about
getting a backup/dump that makes it so you don't have to issue a
separate SQL command.

JD


--
Command Prompt, Inc. - http://www.commandprompt.com/  503-667-4564
PostgreSQL Centered full stack support, consulting and development.
Announcing "I'm offended" is basically telling the world you can't
control your own emotions, so everyone else should do it for you.


Re: per feedback I received directly, pg_dump manpage patch

From
Peter Eisentraut
Date:
On 5/27/15 7:17 PM, Joshua D. Drake wrote:
> The man page does not clearly state that pg_dump will not dump things
> like users or roles. It does mention briefly in the notes section that
> pg_dumpall is needed for items that are set via SET but that isn't very
> clear on the other items.

How about this patch?


Attachment

Re: per feedback I received directly, pg_dump manpage patch

From
Peter Eisentraut
Date:
On 5/27/15 9:31 PM, Alvaro Herrera wrote:
> I didn't check the location of the warning in your patch, or whether a
> warning is really appropriate rather than, say, <note> or a plain
> <para>.

The problem is that every time we get a bug report that some user
tripped over something that was not clear in the documentation, the
patch wants to add a note or warning, which causes some parts of the
documentation to degrade into just a series of warning boxes.  The fewer
warning boxes there are, however, the more likely the user will actually
read all of the documentation, I think.



Re: per feedback I received directly, pg_dump manpage patch

From
"Joshua D. Drake"
Date:
On 05/28/2015 11:16 AM, Peter Eisentraut wrote:
> On 5/27/15 7:17 PM, Joshua D. Drake wrote:
>> The man page does not clearly state that pg_dump will not dump things
>> like users or roles. It does mention briefly in the notes section that
>> pg_dumpall is needed for items that are set via SET but that isn't very
>> clear on the other items.
>
> How about this patch?
>

I would change the word save to backup but other than that, looks good.

JD

>
>
>


--
Command Prompt, Inc. - http://www.commandprompt.com/  503-667-4564
PostgreSQL Centered full stack support, consulting and development.
Announcing "I'm offended" is basically telling the world you can't
control your own emotions, so everyone else should do it for you.


Re: per feedback I received directly, pg_dump manpage patch

From
Peter Eisentraut
Date:
On 5/28/15 3:27 PM, Joshua D. Drake wrote:
>
> On 05/28/2015 11:16 AM, Peter Eisentraut wrote:
>> On 5/27/15 7:17 PM, Joshua D. Drake wrote:
>>> The man page does not clearly state that pg_dump will not dump things
>>> like users or roles. It does mention briefly in the notes section that
>>> pg_dumpall is needed for items that are set via SET but that isn't very
>>> clear on the other items.
>>
>> How about this patch?
>>
>
> I would change the word save to backup but other than that, looks good.

committed