Thread: Some minor error fixes

Some minor error fixes

From
Alexander Law
Date:
Hello,

Please consider fixing some more minor errors.
Patches attached.

Best regards,
Alexander

-----
Alexander Lakhin
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company


Attachment

Re: Some minor error fixes

From
Peter Eisentraut
Date:
On 4/18/16 1:30 AM, Alexander Law wrote:
> Please consider fixing some more minor errors.
> Patches attached.

fixed

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


Re: Some minor error fixes

From
Alexander Law
Date:
Hello Peter,

Thanks!
Please have a look at the some more errors.
Regarding second patch, I think, that inconsistency should be fixed by
omitting '_name', as this parameter could also include a password.

Best regards,
Alexander



04.05.2016 04:10, Peter Eisentraut пишет:
> On 4/18/16 1:30 AM, Alexander Law wrote:
>> Please consider fixing some more minor errors.
>> Patches attached.
>
> fixed
>


Attachment

Re: Some minor error fixes

From
Peter Eisentraut
Date:
On 5/4/16 10:40 AM, Alexander Law wrote:
> Please have a look at the some more errors.
> Regarding second patch, I think, that inconsistency should be fixed by
> omitting '_name', as this parameter could also include a password.

fixed

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


Re: Some minor error fixes

From
Alexander Law
Date:
Thank you!
I have some more errors written down, maybe they are worth fixing too.

Second patch is for consistency in [1].
(I think XMLValidate could be aligned with "IS VALID predicate")
Third patch is for language name teared down in [1].
It seems that root of this typo is as far as in SQL Standard (see M015
description in [2]), but IMO it should be fixed anyway.

[1]
http://www.postgresql.org/docs/9.5/static/unsupported-features-sql-standard.html
[2]
http://jtc1sc32.org/doc/N1501-1550/32N1529T-text%20for%20FCD%209075-9.pdf



04.05.2016 21:11, Peter Eisentraut пишет:
> On 5/4/16 10:40 AM, Alexander Law wrote:
>> Please have a look at the some more errors.
>> Regarding second patch, I think, that inconsistency should be fixed by
>> omitting '_name', as this parameter could also include a password.
>
> fixed
>


Attachment

Re: Some minor error fixes

From
Peter Eisentraut
Date:
On 5/4/16 3:04 PM, Alexander Law wrote:
> Thank you!
> I have some more errors written down, maybe they are worth fixing too.
>
> Second patch is for consistency in [1].
> (I think XMLValidate could be aligned with "IS VALID predicate")
> Third patch is for language name teared down in [1].
> It seems that root of this typo is as far as in SQL Standard (see M015
> description in [2]), but IMO it should be fixed anyway.

I have fixed #1 and #2 but left #3 as it is in the SQL standard.  (I
think M and MUMPS are the same thing.)

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


Re: Some minor error fixes

From
Alexander Law
Date:
Hello Peter,

Thank you!
I see that M is just an alias for MUMPS. I stumbled upon the inconsistency between B015, B115, B125, M022 on the one side and M015 on the other. Anyway, it's definitely not a bug in our docs.

Please look at the following errors/fixes.

Patch #2 is for consistency on [1].
Bug #6 is the most interesting. Table "Table F-17. Hash Algorithm Speeds" on [2] contains following row:
Algorithm | Hashes/sec
crypt-bf/5 | 13504
And there is a following note below the table:
For reference: john -test shows 213 loops/sec for crypt-bf/5. (The very small difference in results is in accordance with the fact that the crypt-bf implementation in pgcrypto is the same one used in John the Ripper.)
It seems that the number 213 is out of sync with the table contents. ("Very small difference" was indeed present before [3]. See [4].)
As I can't reproduce exact numbers on my machine, I suggest to slightly increase the number that was specified in the table (+2 as before).

[1] http://www.postgresql.org/docs/current/static/reference-server.html
[2] http://www.postgresql.org/docs/current/static/pgcrypto.html#PGCRYPTO-HASH-SPEED-TABLE
[3] https://github.com/postgres/postgres/commit/d6464fdc0a591662e5e5ee1b0303932e89cb027c
[4] http://www.postgresql.org/docs/9.1/static/pgcrypto.html#PGCRYPTO-HASH-SPEED-TABLE

Best regards,
Alexander


14.05.2016 04:41, Peter Eisentraut пишет:
On 5/4/16 3:04 PM, Alexander Law wrote:
Thank you!
I have some more errors written down, maybe they are worth fixing too.

Second patch is for consistency in [1].
(I think XMLValidate could be aligned with "IS VALID predicate")
Third patch is for language name teared down in [1].
It seems that root of this typo is as far as in SQL Standard (see M015
description in [2]), but IMO it should be fixed anyway.

I have fixed #1 and #2 but left #3 as it is in the SQL standard.  (I think M and MUMPS are the same thing.)


Re: Some minor error fixes

From
Alexander Law
Date:
Hello Peter,

Thank you!
I see that M is just an alias for MUMPS. I stumbled upon the inconsistency between B015, B115, B125, M022 on the one side and M015 on the other. Anyway, it's definitely not a bug in our docs.

Please look at the following errors/fixes.

Patch #2 is for consistency on [1].
Bug #6 is the most interesting. Table "Table F-17. Hash Algorithm Speeds" on [2] contains following row:
Algorithm | Hashes/sec
crypt-bf/5 | 13504
And there is a following note below the table:
For reference: john -test shows 213 loops/sec for crypt-bf/5. (The very small difference in results is in accordance with the fact that the crypt-bf implementation in pgcrypto is the same one used in John the Ripper.)
It seems that the number 213 is out of sync with the table contents. ("Very small difference" was indeed present before [3]. See [4].)
As I can't reproduce exact numbers on my machine, I suggest to slightly increase the number that was specified in the table (+2 as before).

[1] http://www.postgresql.org/docs/current/static/reference-server.html
[2] http://www.postgresql.org/docs/current/static/pgcrypto.html#PGCRYPTO-HASH-SPEED-TABLE
[3] https://github.com/postgres/postgres/commit/d6464fdc0a591662e5e5ee1b0303932e89cb027c
[4] http://www.postgresql.org/docs/9.1/static/pgcrypto.html#PGCRYPTO-HASH-SPEED-TABLE

Best regards,
Alexander


14.05.2016 04:41, Peter Eisentraut пишет:
On 5/4/16 3:04 PM, Alexander Law wrote:
Thank you!
I have some more errors written down, maybe they are worth fixing too.

Second patch is for consistency in [1].
(I think XMLValidate could be aligned with "IS VALID predicate")
Third patch is for language name teared down in [1].
It seems that root of this typo is as far as in SQL Standard (see M015
description in [2]), but IMO it should be fixed anyway.

I have fixed #1 and #2 but left #3 as it is in the SQL standard.  (I think M and MUMPS are the same thing.)


Attachment

Re: Some minor error fixes

From
Peter Eisentraut
Date:
On 5/14/16 2:23 AM, Alexander Law wrote:
> Please look at the following errors/fixes.

I've applied 2, 3, 4, 5.

1 was correct according to my math. (115.9-9.8)/100000*1000*1000 = 1061

6 looked too complicated to me. ;-)  Can you explain where you got your
number from?

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


Re: Some minor error fixes

From
Alexander Law
Date:
Thank you, Peter.

Regarding 1, you're right, I didn't see "per row" in that sentence and
decided that it was total overhead (and then again I should change nsec
to msec).

Regarding 6, please look at the old documentation:
http://www.postgresql.org/docs/9.1/static/pgcrypto.html#PGCRYPTO-HASH-SPEED-TABLE
It contains "crypt-bf/5 | 211" in the table and "john -test shows 213
loops/sec for crypt-bf/5" below the table. (The difference is 2 loops
per second).

Current documentation:
https://www.postgresql.org/docs/current/static/pgcrypto.html#PGCRYPTO-HASH-SPEED-TABLE
contains:
"crypt-bf/5     13504" in the table (number is increased with the faster
CPU) and still "john -test shows 213 loops/sec for crypt-bf/5" below the
table.
So I propose to change 213 below the table to 13504 + 2 (previous
difference).
Or maybe we should rerun all the benchmarks and update all the numbers
(see commit d6464fdc).

Best wishes,
Alexander


15.07.2016 05:36, Peter Eisentraut пишет:
> On 5/14/16 2:23 AM, Alexander Law wrote:
>> Please look at the following errors/fixes.
> I've applied 2, 3, 4, 5.
>
> 1 was correct according to my math. (115.9-9.8)/100000*1000*1000 = 1061
>
> 6 looked too complicated to me. ;-)  Can you explain where you got your
> number from?
>



Re: Some minor error fixes

From
Peter Eisentraut
Date:
On 7/15/16 12:06 AM, Alexander Law wrote:
> Regarding 6, please look at the old documentation:
> http://www.postgresql.org/docs/9.1/static/pgcrypto.html#PGCRYPTO-HASH-SPEED-TABLE
> It contains "crypt-bf/5 | 211" in the table and "john -test shows 213
> loops/sec for crypt-bf/5" below the table. (The difference is 2 loops
> per second).
>
> Current documentation:
> https://www.postgresql.org/docs/current/static/pgcrypto.html#PGCRYPTO-HASH-SPEED-TABLE
> contains:
> "crypt-bf/5     13504" in the table (number is increased with the faster
> CPU) and still "john -test shows 213 loops/sec for crypt-bf/5" below the
> table.
> So I propose to change 213 below the table to 13504 + 2 (previous
> difference).

makes sense, fixed

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