Talk:PostgreSQL

Upcoming features

Copied from [1] WP:CRYSTAL doesn't apply to the near future (with a set date), at least (notable) movies. [It's a question if next future releases of PostgreSQL is notable..] I guess planned features can be cancelled (at any time?). In case they are in beta I would say including is ok. As the upcoming release (as all are) is open source it is released in some sense (and could be used, but not recommended for production, while testing is ok and recommended). Really all committed code is release (but not notable). Where should we draw the line (believe will end up in next release), beta, release candidate, alpha?.

Unverified article linked

"Matloob Khushi performed benchmarking between PostgreSQL 9.0 and MySQL 5.6.15 for their ability to process genomic data. In his performance analysis he found that PostgreSQL extracts overlapping genomic regions eight times faster than MySQL using two datasets of 80,000 each forming random human DNA regions. Insertion and data uploads in PostgreSQL were also better, although general searching ability of both databases was almost equivalent"

This above sequence and linked article (i cannot even say work), is totally unverified. It do not contain any code, sample data, description of algorithms, server setups, configurations, simply nothing. It is unproveable. There is only something like postgres is 1200 x faster in this and in that. 178.235.4.27 (talk) 18:59, 16 October 2022 (UTC)

I suspect you are basing your claims on having read only the abstract of the article. The full article has full details of all matters you dispute. Something being behind a paywall doesn't mean it can't be used as a source. As well, there are alternative means of accessing the full article. I'm unwilling to go into detail as there are assorted legal aspects to it, and I don't know whether there are wikipedia policies at work, however, I can certainly wikilink to Sci-Hub. cheers. anastrophe, an editor he is. 20:57, 16 October 2022 (UTC)

Where is the full article an why not linked? 178.235.4.27 (talk) 14:18, 17 October 2022 (UTC)

Please reread what I just wrote, and do so carefully. The answer to your question is there. Editors are expected to put some effort into researching their own inquiries. cheers. anastrophe, an editor he is. 18:43, 17 October 2022 (UTC)

I have read it carefully, but I have not found the full article anywhere or the possibility to buy it ;-)

I have contacted the university but they refuse to comment. It really does look like pseudoscience.

I have "removed" this year 11 psudoscience articles from the web.

If you read this shortcut of the article thoroughly, you will see, for example, that he writes that he has tested many databases and in the next sentence that postgres is faster than mysql.

So he finally tested it on several databases or was he just comparing postgres with mysql? 178.235.4.27 (talk) 21:22, 18 October 2022 (UTC)

"I have 'removed' this year 11 psudoscience articles from the web." Can you please explain what you mean by "removed"? It's rather silly to deride the study without having read it.

At the risk of sanctions (again, I'm not sure of policy here), here is what you do:

1. Click on the reference link in this article to the abstract of Matloob Khushi's study that you are concerned with here.
2. Note that a few lines below the title, it lists a "PMID". Copy that number.
3. Go to the Wikipedia article entitled Sci-Hub
4. Visit the URL that is helpfully provided in the infobox.
5. At the destination, immediately under the large letters saying sci-hub, there is a link labeled back to main. Click it.
6. In the large Enter your reference box, paste the PMID number you copied as above. Click the "open" button.
7. Read the actual full article.
8. cheers. anastrophe, an editor he is. 21:42, 18 October 2022 (UTC)

Thank you. But i have followed your points and at the end i see same article as before only a better formated.

But still without any info to reporduce anything ;-)

Still pseudoscience. I can write same article and write there that Firebird database outperform PostgreSQL and it run genomic comarision algorithm named MySuperHiperAlgorithm 1200x faster ;-)

How can you prove that it is not true without any details? Any data, config, DDL, DML, ... 178.235.4.27 (talk) 13:08, 19 October 2022 (UTC)

By "I have 'removed' 11 psudoscience articles from the web" i mean, I wrote to the universities about the essays. I managed to convince universities that this was psudoscience and they removed it. Unfortunately, some universities are not interested, they probably get more grants the more articles they have ;-) 178.235.4.27 (talk) 13:14, 19 October 2022 (UTC)

So you still only see the abstract? In that case, I'm unable to help further, as the full article is available with all details. Perhaps write to the article author for more guidance. However, until you read the full article, rather than just the abstract, do not remove the section from the article, as the information is verifiable. cheers. anastrophe, an editor he is. 17:38, 19 October 2022 (UTC)

Does it mean that by Sci-Hub you have access to full article? How many pages it have? 178.235.4.27 (talk) 10:12, 21 October 2022 (UTC)

Describing PG as relational v.s. object-relational in the lead

tl;dr Calling PG object-relational is not useful.

Hi,

I believe that mention of object-relational databases in the lead text is not desireable. Both relational and object-relational databases are theoretical in that there are no actual implementations of either. Nobody really talks about using Postgres as an object-relational database any more than they talk of using other databases, like MySQL, as an object-relational db. Even though these other databases have features found in object-relational dbs, to roughly the same degree as such features are found in Postgres. The industry as a whole categorizes databases broadly into relational dbs and NoSQL dbs, and only secondarily into navigational/hierarchical, relational, and key-value stores with hardly any mention these days of object-relational dbs. All the concepts in the lead text should be at least marginally recognizable to a moderately sophisticated reader. To my mind object-relational database-ness does not meet this criteria.

I don't believe that the lead text needs to get into the weeds when it comes to detailing exactly what theoretical models are supported to what degree. Postgres is used as, and mostly is, a relational db. In real object-relational dbs, to my mind, The ORDBMS (like ODBMS or OODBMS) is integrated with an object-oriented programming language ... [having] program objects [which] must be storable and transportable for database processing, therefore they usually are named as persistent objects. There is no integration in Postgres with any object-oriented programming language, beyond the ability to embed arbitrary languages within Postgres, and no provision for persistent object stores. Obviously, such features can be layered on top of Postgres (e.g. SQLAlchemy), but this can be said of any relational db. For these reasons I don't believe Postgres qualifies as a fully object-relational db. From a practical perspective an object-relational db should fully support object persistence in at least some object-oriented language or it does not qualify.

My theory is that the PG home page says "object-relational" because much of the original development happened in the 90's when "object-relational" was all the rage. That was when many object-relational-related features, like type extensions and inheritance and table inheritance were introduced into Postgres. Calling Postgres object-relational was a way to hilight the then-advanced features that Postgres sported.

Despite sporting some object-relational features I don't believe Postgres is really object-relational. It is not used in an object-relational fashion, at least not any more than other modern popular relational databases. And because regular relational databases can all now be used with object-oriented programming languages by way of an ORM mapper the industry no longer talks much, if at all, about object-relational databases. Given these considerations, and especially because object-relational is not a widespread term-of-art, it is not useful to use "object-relational" in the lead text. — Preceding unsigned comment added by Kop (talk • contribs) 23:07, 24 December 2023 (UTC)

Hi hi! Great points, as I read this I wonder if this is a useful historical aspect of the design of Postgres to mention. There is a source referenced #52 that references an object relational mapping which it appears most web frameworks offer instead, but I'm curious about how this design may have supported the adoption of Postgres (and perhaps why they still brand it as such on the main site). An actual submission is better than a suggestion, I know, but I trust your point of view on this @Kop! 2600:1700:5040:5150:FC62:C46C:7608:4D77 (talk) 16:59, 11 January 2024 (UTC)

References

Initial release of 8 July

I believe this is wrong even though the team celebrated its 12 year birthday that day. Initial release was in 1997. The article itself says this "The first PostgreSQL release formed version 6.0 on January 29, 1997". 213.131.36.174 (talk) 17:24, 10 June 2024 (UTC)

Standards compliance

The standards compliance section says 'PostgreSQL conforms to at least 170 of the 177 mandatory features for SQL:2023 Core conformance", and no other databases fully conformed to it'. However, Mimer SQL seems to support all core features, see https://download.mimer.com/pub/developer/standard/Core-SQL-Feature-Summary.pdf. Or do I misunderstand this? — Preceding unsigned comment added by Nowthenlater (talk • contribs) 19:34, 22 October 2024 (UTC)