Re: [HACKERS] pgindent weirdnessf

2011-04-20 Thread Andrew Dunstan 



On 04/20/2011 01:59 PM, Bruce Momjian wrote:

I do, agree, though, it would be nice to find out what changed that
caused this.



I am 100% certain that it's the tools that have changed. I bet if I were 
to hunt in my pile of old DVDs and find installation media for Fedora 6 
or thereabouts and set it up on a VM I'd be able to reproduce the old 
list. But it would be a serious waste of my tolerably precious time.


cheers

andrew

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] pgindent weirdnessf

2011-04-20 Thread Bruce Momjian 
Robert Haas wrote:
> On Wed, Apr 20, 2011 at 12:33 PM, Andrew Dunstan  wrote:
> > You can contribute to the list by running a buildfarm animal on your machine
> > and running its find_typedefs occasionally. This is not just about me. I
> > have asked on numerous occasions for other people to contribute, and the
> > response has been deafening silence. The ?reason we got to this place is
> > that people complained that your list was insufficiently complete, so I
> > added a facility for buildfarm animals to generate their own lists, so we
> > could get wider platform coverage. So my response to anyone who says "well,
> > it works on my box" is "then why isn't your box doing it for the buildfarm?"
> 
> This is all well and good up to a point, but if Bruce's ancient BSDi
> machine is the only one that can properly find these symbols, then we
> are courting disaster by relying on it, even if he does run a
> buildfarm animal there.  I can't help thinking there must be some
> other explanation for this change.

Uh, just a reality check, but our "courting disaster" means we will have
an extra space after some asterisks in the source code.  ;-)

I do, agree, though, it would be nice to find out what changed that
caused this.

-- 
  Bruce Momjian  http://momjian.us
  EnterpriseDB http://enterprisedb.com

  + It's impossible for everything to be true. +

-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] pgindent weirdnessf

2011-04-20 Thread Robert Haas 
On Wed, Apr 20, 2011 at 12:33 PM, Andrew Dunstan  wrote:
> You can contribute to the list by running a buildfarm animal on your machine
> and running its find_typedefs occasionally. This is not just about me. I
> have asked on numerous occasions for other people to contribute, and the
> response has been deafening silence. The  reason we got to this place is
> that people complained that your list was insufficiently complete, so I
> added a facility for buildfarm animals to generate their own lists, so we
> could get wider platform coverage. So my response to anyone who says "well,
> it works on my box" is "then why isn't your box doing it for the buildfarm?"

This is all well and good up to a point, but if Bruce's ancient BSDi
machine is the only one that can properly find these symbols, then we
are courting disaster by relying on it, even if he does run a
buildfarm animal there.  I can't help thinking there must be some
other explanation for this change.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] pgindent weirdnessf

2011-04-20 Thread Andrew Dunstan 



On 04/20/2011 11:48 AM, Bruce Momjian wrote:

I assume you are using -g, right?



Of course I did. I wouldn't have any symbols at all if I didn't.


The BulkInsertStateData typedef looks pretty normal:

typedef struct BulkInsertStateData
{
BufferAccessStrategy strategy;  /* our BULKWRITE strategy 
object */
Buffer  current_buf;/* current insertion target page */
}   BulkInsertStateData;

I tested my BSD machine using src/tools/find_typedefs and it does show
BulkInsertStateData.



You can contribute to the list by running a buildfarm animal on your 
machine and running its find_typedefs occasionally. This is not just 
about me. I have asked on numerous occasions for other people to 
contribute, and the response has been deafening silence. The  reason we 
got to this place is that people complained that your list was 
insufficiently complete, so I added a facility for buildfarm animals to 
generate their own lists, so we could get wider platform coverage. So my 
response to anyone who says "well, it works on my box" is "then why 
isn't your box doing it for the buildfarm?"


cheers

andrew

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] pgindent weirdnessf

2011-04-20 Thread Bruce Momjian 
Andrew Dunstan wrote:
> It's tagged as a structure type by objdump, but not as a typedef:
> 
> <1><40055>: Abbrev Number: 4 (DW_TAG_typedef)
> <40056>   DW_AT_name: (indirect string, offset: 0x6bf6):
> BulkInsertState
> <4005a>   DW_AT_decl_file   : 30
> <4005b>   DW_AT_decl_line   : 32
> <4005c>   DW_AT_type: <0x40060>
> <1><40060>: Abbrev Number: 7 (DW_TAG_pointer_type)
> <40061>   DW_AT_byte_size   : 8
> <40062>   DW_AT_type: <0x40066>
> <1><40066>: Abbrev Number: 13 (DW_TAG_structure_type)
> <40067>   DW_AT_name: (indirect string, offset: 0x66bf):
> BulkInsertStateData
> <4006b>   DW_AT_byte_size   : 16
> <4006c>   DW_AT_decl_file   : 31
> <4006d>   DW_AT_decl_line   : 30
> <4006e>   DW_AT_sibling : <0x4008b>
> 
> I can pull out those too if you want them in the list, but it would 
> possibly add a LOT of names to the list.
> 
> I did carefully warn you about the need to check the effects of the 
> changes when I committed the new list.
> 
> It looks like quite a few of the deletions come into this category, for 
> example just looking at the diff here 
> 
>  
> I see AggHashEntryData, AggStatePerAggData, AllocBlockData, and 
> AllocChunkData from among the first few that were deleted and all are in 
> the same category.
> 
> I wondered if this is some sort of optimizer effect, but building with 
> -O0 doesn't seem to affect it.

I assume you are using -g, right?

The BulkInsertStateData typedef looks pretty normal:

typedef struct BulkInsertStateData
{
BufferAccessStrategy strategy;  /* our BULKWRITE strategy 
object */
Buffer  current_buf;/* current insertion target page */
}   BulkInsertStateData;

I tested my BSD machine using src/tools/find_typedefs and it does show
BulkInsertStateData.

-- 
  Bruce Momjian  http://momjian.us
  EnterpriseDB http://enterprisedb.com

  + It's impossible for everything to be true. +

-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers