[ CC'ed to docs. Hope that is OK.]
These are excellent points. I have merged items 3.5 and 3.6 as you
suggested. They clearly should have been one item.
As far as disk computation, you are correct also. I improperly
estimated the size of integers in that file. I have created a more
typical file of one integer and one 20-char length string and used that
in the example. I also changed the text to say "up to five times"
rather than "six times". My new example shows roughly three times
increase so I think "up to five times" is much more accurate.
I have updated CVS and the web page so you can see my exact changes.
Thanks for the assistance.
---------------------------------------------------------------------------
Viktor Vislobokov wrote:
> Hello Bruce!
>
> At first, I'm sorry my bad English.
>
> I'm like PostgreSQL and I want to make little contribution to
> this great work.
> I translated PostgreSQL FAQ in Russian. My translation
> based on PostgreSQL 7.2 distribution for RawHide Linux (Red Hat clone).
> Can I send it to you, that you place it into official distribution?
> I'm not professional translator, but I try to make quality
> translation.
>
> After translation process, I have some questions and I
> want to discuss it with you.
>
>
> What you think about 3.5 and 3.6 questions?
> I think, answers are duplicate. May be need
> merge it?
>
> I read answer at question 4.6.
> I not agree with it, because the data is incorrect.
>
> You are talking about 300,000 lines, 2 integer at line
> and 2.4Mb total size in order text file. But
>
> 2.4*1024*1024
> ------------- = 8 bytes at line
> 300,000
>
>
> Let line containts two integers
>
> 65535 22767
>
> In this case - 10 digits plus space plus end line character = 12 bytes
>
> However, PostgreSQL INT type can store 4th bytes interger, i.e.
> any integer from 0 to 0xFFFFFFFF in hex and from 0 to 4294967295 in
> decimal.Thus, if a line in text file contains, for example
>
> 4294967295 4294967295
>
> total size of this line will be 22 bytes!
>
> Finally, result is strongly depends of integer's values!
> IMHO, If numbers will be large, database size can be even LESS
> text file size.
>
> With best wishes, Victor Vislobokov
> Perm, Russia
>
>
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026
