Thread

Commits

  1. Remove error message hints mentioning configure options

  1. Remove error message hints mentioning configure options

    Peter Eisentraut <peter.eisentraut@enterprisedb.com> — 2022-04-07T07:19:14Z

    We have some error messages like this:
    
    SELECT xml_is_well_formed('<abc/>');
    ERROR:  unsupported XML feature
    DETAIL:  This functionality requires the server to be built with libxml 
    support.
    HINT:  You need to rebuild PostgreSQL using --with-libxml.
    
    This patch removes these kinds of hints.
    
    I think these hints are usually not useful since users will use packaged 
    distributions and won't be interested in rebuilding their installation 
    from source.  Also, we have only used these kinds of hints for some 
    features and in some places, not consistently throughout.  And of course 
    there are build systems that don't use configure.  The information 
    "needs to be built with XXX" or "was not built with XXX" should be 
    enough for those interested in actually changing their build 
    configuration to figure out what to do.
  2. Re: Remove error message hints mentioning configure options

    Andres Freund <andres@anarazel.de> — 2022-04-07T07:34:31Z

    Jo.
    
    On 2022-04-07 09:19:14 +0200, Peter Eisentraut wrote:
    > I think these hints are usually not useful since users will use packaged
    > distributions and won't be interested in rebuilding their installation from
    > source.  Also, we have only used these kinds of hints for some features and
    > in some places, not consistently throughout.  And of course there are build
    > systems that don't use configure.  The information "needs to be built with
    > XXX" or "was not built with XXX" should be enough for those interested in
    > actually changing their build configuration to figure out what to do.
    
    +1
    
    > diff --git a/src/test/regress/expected/compression_1.out b/src/test/regress/expected/compression_1.out
    > index 1ce2962d55..c0a47646eb 100644
    > --- a/src/test/regress/expected/compression_1.out
    > +++ b/src/test/regress/expected/compression_1.out
    
    The xml stuff is at least old, but compression_1 isn't. I think we've tried to
    avoid long alternative output files where everything fails due to being
    unsupported for a while now? Robert, Dilip? See uses of :skip_test in various
    tests.
    
    Greetings,
    
    Andres Freund
    
    
    
    
  3. Re: Remove error message hints mentioning configure options

    Daniel Gustafsson <daniel@yesql.se> — 2022-04-07T09:25:10Z

    > On 7 Apr 2022, at 09:19, Peter Eisentraut <peter.eisentraut@enterprisedb.com> wrote:
    > 
    > We have some error messages like this:
    > 
    > SELECT xml_is_well_formed('<abc/>');
    > ERROR:  unsupported XML feature
    > DETAIL:  This functionality requires the server to be built with libxml support.
    > HINT:  You need to rebuild PostgreSQL using --with-libxml.
    > 
    > This patch removes these kinds of hints.
    > 
    > I think these hints are usually not useful since users will use packaged distributions and won't be interested in rebuilding their installation from source.
    
    Agreed, +1 on this patch.  Grepping the code I was also unable to find any
    other user-facing instances.
    
    --
    Daniel Gustafsson		https://vmware.com/
    
    
    
    
    
  4. Re: Remove error message hints mentioning configure options

    Tom Lane <tgl@sss.pgh.pa.us> — 2022-04-07T13:49:01Z

    Andres Freund <andres@anarazel.de> writes:
    > On 2022-04-07 09:19:14 +0200, Peter Eisentraut wrote:
    >> I think these hints are usually not useful since users will use packaged
    >> distributions and won't be interested in rebuilding their installation from
    >> source.
    
    > +1
    
    +1, those hints are from another era.
    
    > The xml stuff is at least old, but compression_1 isn't. I think we've tried to
    > avoid long alternative output files where everything fails due to being
    > unsupported for a while now? Robert, Dilip? See uses of :skip_test in various
    > tests.
    
    Agreed, we have a better technology now for stub test results.  Come
    to think of it, xml.sql really ought to be redone along the same lines.
    
    			regards, tom lane