Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Problems with Oracle SQL

codingtofreedom.com

Post reply on HN
Print view View original

261–270 of 277 posts

Re: Problems with Oracle SQL

#261

Post by puppet-master »

segfaultbuserr wrote: post #258
puppet-master wrote: post #256

Earlier quoted context omitted.

> defrag.exe simply would not run no matter what (I needed it to TRIM the SSD) Would be interested in the logic here, given that it runs counter to what I thought was extremely obvious advice by now. Defrag generates writes to rearrange blocks in the _virtual_ view of storage visible to the OS, but even after defragmentation, the _physical_ placement is totally outside the control of anything except the firmware. If…

> All you can do is let it know that some range of sectors is not used, which is what TRIM enables, and that does not require defragmentation upfront. This is exactly what defrag.exe does on an SSD today. Since Windows 10 (or 8?), instead of doing a "real" defrag, defrag.exe also has an option to issue TRIM commands to unused blocks on an SSD. In other words, it works like fstrim(1). The time has changed and you must…

Thanks for the excellent explanation, but the attitude and needless personalization was unnecessary. This functionality was previously available via the Optimize-Volume cmdlet, I guess it makes sense the defragmenter was updated with similar functionality to account for the typical Windows user.

Re: Problems with Oracle SQL

#262

Post by marcosdumay »

kaba0 wrote: post #253
marcosdumay wrote: post #76

Earlier quoted context omitted.

> Oracle it not that bad anyway, it's just stupidly expensive. Full of bugs, some that destroy speed, some that destroy data, some that just make a feature that you need unusable. More heavyweight than lead (although it gets fast after you throw enough hardware). Lacking any capable or usable management interface (but then, that excludes everyone except for postgres and mysql). Impossible to program. Impossible to pr…

Any source on the bugs?

The Oracle's documentation on bugs is paywalled... What is actually an improvement over others like Microsoft, that denies a bug with all their might, up to the point it is fixed, and then say the minimum possible.

Anyway the most recent pair I found on the wild was a problem that made indexes of georeferenced data fail at random, pushing your queries into a non-indexed search and breaking things like materialized views, and one that causes some inconsistency on testing clobs for null or empty string (what made them impossible to test for either when it applies). But there's a well known one that breaks optimization plans at random and goes with the last try (it doesn't matter if that it will scan that table 100000 rows 1000 times), and there's some bug where if you create some tables, populate them, drop them and repeat enough times you will lose your database. But that's just from the top of my head.

Oh, of course, that isn't including designed behaviors like the one that takes your database offline if you don't do backups often enough (where "often enough" is something that you can estimate but never be sure about its frequency).

Re: Problems with Oracle SQL

#263

Post by metadata »

chasil wrote: post #223
metadata wrote: post #45

Earlier quoted context omitted.

Postgres has Oracle layer, available as a commercial variant by EnterpriseDB. I don't think MariaDB implementation can compete, but EDB Postgres is solid. Basically a native implementation of Oracle in Postgres.

The Oracle layer was also added to IBM DB2. Which one? Good question. The mainframe, as/400, and Windows versions are all on separate source code last I heard. https://www.enterprisedb.com/news/enterprisedb-and-ibmr-coll...

DB2 LUW (Linux/Unix/Windows), DB2 iSeries, and DB2 z/OS are three different databases. With the same name - I guess for branding.

Re: Problems with Oracle SQL

#264

Post by metadata »

garenp wrote: post #77
metadata wrote: post #45

Earlier quoted context omitted.

Postgres has Oracle layer, available as a commercial variant by EnterpriseDB. I don't think MariaDB implementation can compete, but EDB Postgres is solid. Basically a native implementation of Oracle in Postgres.

Does it replicate the same kind of "evil" bugs as in the article, or does it provide an improvement?

AFAIK it works exactly as Oracle. Doing something barely similar would probably introduce some very expensive bugs in production.

Re: Problems with Oracle SQL

#265

Post by cultofmetatron »

edgriebel wrote: post #86
ok123456 wrote: post #61

Earlier quoted context omitted.

No. Some technologies are hostile to the people forced to use them.

That's pretty much what OC is saying, any technology good enough to be commonly used is going to suck and/or have sharp edges

to a degree thats true but the amount of vitriol expressed to oracle seems a bit warranted. I'm a user of postgres. its widely supported on cloud offerings. Every startup I worked in used it or mysql so it could be argued that postgres is used in a lot more places.

While I certainly have a few issues with postgres, It does the job with minimal bullshit and most of the time, just works. I'm at a point where I just trust it to do its damn job. How often do you hear people complain about postgres or mysql the way you hear people complain about oracle's enterprise db offerrings?

Re: Problems with Oracle SQL

#266

Post by infamia »

jiggawatts wrote: post #249
infamia wrote: post #232

Earlier quoted context omitted.

PowerShell elegant? Wow. No way. I've written hundreds of thousands of lines of PowerShell on Windows over the years. PowerShell is great for one-liners and very short scripts. It also has many cool hooks into the OS that makes things more comfy. That said, as soon as you need to write a modestly long script or need something remotely performant, you're going to want to reach for the nearest pencil to shove into your…

An observation I've made over decades of development in dozens of languages on many platforms is that humans often confuse frequency for speed. They assume that because they are making "many actions per unit time" that they are getting to their solution goal faster. This is most obviously noticeable (to me at least), when debating ergonomics with people that prefer UNIX platforms, especially bash and text-based confi…

> Verbose commands are an enabler. They enable novice users to read scripts, instead of only masters being able to write them. Long, systematically and consistently named commands enable discovery through wildcard searches.

For short scripts, this works great and improves discoverability for newcomers. This is the siren's song of PowerShell. However, the long commands and particularly the often unneeded/overly verbose parameters frequently creates a wall of text. This really hurts readability for everyone in all but the shortest scripts. Additionally, this wall-of-text that is all to common in PowerShell scripts is very intimidating to newcomers.

Regarding Tab completion in PowerShell, it frequently isn't terribly helpful. To work yourself up to something like Get-ItemPropertyValue is quite an incantation to remember, while being worse than a lot of bash ergonomics.

> have you considered writing "heavyweight" PowerShell modules in C#?

That is an interesting approach and would make life more bearable in MS-only shops. However, I can do the same thing with Python, which is superior to PS in so many ways and doesn't suffer any huge, glaring flaws (as PS does).

Re: Problems with Oracle SQL

#267

Post by johnthescott »

tracker1 wrote: post #119
zabzonk wrote: post #55

Earlier quoted context omitted.

> so the choice was Oracle or DB2 Sybase, Ingres

Worth noting, that Microsoft SQL Server started life as a Sybase fork.

not so sure that is correct. ms sql WAS sybase from 1989 until the fork in 1993 and i have been told that substantial changes to the codebase did not occur until years after the 1993, during the jim gray tenure.

Re: Problems with Oracle SQL

#268

Post by icedchai »

marcosdumay wrote: post #233
icedchai wrote: post #93

Earlier quoted context omitted.

varchar1 was varchar. I should say "is", since varchar is still supported, they just recommend varchar2 since varchar "might change" (probably not...)

That one actually changed already. If you follow the new recommendations, varchar is as good as varchar2.

Good to know! I haven't had to touch an Oracle database in almost 15 years, but one never knows.

Re: Problems with Oracle SQL

#269

Post by laumars »

revscat wrote: post #117
laumars wrote: post #113

Earlier quoted context omitted.

It’s more than that. Oracle is sold off the back of its middleware applications. Which equally suck too…but at least they’re solving difficult problems that most other database vendors don’t both with.

Such as?

Such as enterprise level accountancy solutions, human resource packages, payroll solutions (which is the where I worked with Oracle) and so on and so forth.

A lot of the time when businesses by Oracle, they’re buying into the middleware rather than the RDBMS.

Re: Problems with Oracle SQL

#270

Post by simonw »

jiggawatts wrote: post #225
simonw wrote: post #118

There was a comment on here a few years ago from a developer with experience working on the Oracle database codebase which put me off it for life: https://news.ycombinator.com/item?id=18442941

To me, millions test sounded like a good thing. The main issue I had with the description of that workflow is that if the tests take 20-30 hours to run, someone in Oracle is not spending sufficient money on build hardware! They should have tens of thousands of servers, not hundreds, and they should be kitted out with the fastest processors and NVMe drives money can buy. If you look at how companies like NVIDIA do thi…

It's possible to have millions of tests and NOT have the problems described in that story. SQLite has a famously thorough test suite and I've seen the SQLite developers turn around a suggested improvement to their codebase - including authoring the fix, testing it and landing it, in just a few hours.
Post reply on HN