You can use the REGEXP_REPLACE() function with no additional flags.
select regexp_replace('1. xxxxxx1. xxxx1. ', '1. ', '');
returns:
xxxxxx1. xxxx1.
db<>fiddle here
When supplying a link that highlights the text on a web page (in Chrome, see note below), a tilde character ~ is used as part of the URL to identify text to highlight:
https://meta.stackexchange.com/q/370143/#:~:text=highlights%20the%20text%20on%20a%20web%20page
When using a link like this with...
When the text you want to highlight appears in multiple places, that allows you to specify what comes immediately before and after the text you meant to highlight.
@McNets - take a look here - our old pal, the array question by your neighbour - can't find the post where I dug it up, but it turns 34 lines of code into 4! :-)
@McNets - it's not for any upvotes - I just think it's a pretty nifty piece of code and I thought that since you put a bit of time into it, you'd be interested. AFAIC, there's more to life than unicorn points... lernin' (to spell for example) and stuff! :-)
@PaulWhite Now I'm getting flashbacks of my endless battles with underperforming SANs in the 2000's. For a first approximation, SANs are 10x the price per GB and 100x the price per IOPS of direct attach storage.
Lovin me some NVMe drives these days. File copies at 1.5GB/sec.
@ConcernedOfTunbridgeWells I'd be more interested if it could be done in SQL! Take a look at explainextended.com for some weird and wonderful things you can do in SQL, but thanks for the heads up.
@PaulWhite I have an Evo 980 waiting to go into something that's rated at 7GB/sec read on a pci-e v4.0 interface. The system disk in my machine is an older P750, really a 1st gen NVMe and the data disk is an S4601, which is for 'read intensive' workloads and doesn't have spectularly good write performance.
Still, it's pretty quick. You can 7.68TB 3D QLC drives for about £600 these days.
I feel that all that time I put into fibre-channel drives was pretty much wasted these days.
In 10 years time when everything will be on single-byte addressable, non-volatile RAM (will make NVMe look like punched-card readers), we'll laugh at these numbers.
@PaulWhite Intel's bad, I think - as far as I can tell, Optane is some sort of low-density, high speed/endurance flash with a battery-backed DRAM cache. However, they did a fantastic job of obfuscating what it actually was and did, so I think they got ignored by the industry as nobody had a clue why they might want to use it.
Yeah I think it made sense when people were still using spinning drives and maybe some old slower SSDs. Fast NVMe is so cheap now. Maybe there's still a place for Optane but I'm not aware of it
If Intel had at least said 'Put tempdb on this.', folks might have actually used it.
I think their Optane RAM use case for non-volatile VM image caches sort of makes sense if you want to spin them up quickly for some reason. However, I think even in the wonderful world of cloud computing, that's still a pretty niche use case.
And Optane is very expensive - A 750Gb Optane drive is £1,000 or more on the street, let alone what it might cost through a vendor. For that money you can get 15 or 30TB of NVME flash.
One thing I will give Optane is that it has very good endurance by flash standards.
I've still got the machine with the old IOdrive 1 in storage somewhere. Sadly, it's way too old for Windows 11 support so I think it might be destined for recycling at some point.
Even my current Z440 has a CPU generation too old for Windows 11.
Pity, really, it's a perfectly good machine. Its only vice is the NVMe drives; they occasionally glitch in POST on boot-up with a pci-e error.
@ConcernedOfTunbridgeWells Yes my W541 is the same, bit older though. I could run W11 now but I choose not to. Seems too early and I'm quite happy on W10.
I've got a P52s to replace the old W520, although as I've not been in an office since March last year, it hasn't had a lot of use.
Fortunately that's recent enough for Windows 11 support. I'm thinking of trying Windows Server 2019 out on the Z440, as that has extended support until 2029.
In the past I've had good results running server versions as a desktop.
@Vérace-getVACCINATEDNOW One can get a lot of computrons into a laptop these days.
I feel most of my operation could be serverless now. Not because of cloud services, but because I can get a PC powerful enough to run a pretty major workload off ebay for less than a month's rent on a hosted server of equivalent spec.
The only server I have at the moment is a NAS.
@Vérace-getVACCINATEDNOW Back when spinnies were the state of the art and machines had memory capacity measured in megabytes, there was a potential conflict between logs and data disks.
@Vérace-getVACCINATEDNOW Log writes were mostly small and sequential, and were relatively fast. Data disk writes were random access. You put them on separate disks so the random access data traffic didn't turn the sequential log traffic into random access traffic, which would cause a bottleneck on the logs. It would make a difference for direct attach storage, less so for SAN storage with a battery backed cache, and little if anything on flash storage.
Although you might not notice it on the SAN as everything is on shared volumes and pretty slow at the best of times.
@ConcernedOfTunbridgeWells My holiday spare time will be spent configuring the machine - major geekfest! I may surface from time to time for food, otherwise, it's the mandate for me! Yaay...
$%^&-ing tablet is correcting my spewing again - I didn't make this up - it turned spelling into spewing!
@ConcernedOfTunbridgeWells Do you still see people separating log/data even though in almost all cases it is based on information that is ~20 years out of date?
@PaulWhite I specified separate log drives on some old G5 vintage servers that were repurposed for development machines around 2012. These had 6 spinny disks with local host-based RAID in them.
Everything I've worked with since then has been either all-flash, set up by somebody else or I've been too plebby to get involved at that level. I'm not sure.
@Vérace-getVACCINATEDNOW My development builds used to take 3 days to install and patch at one point. It's a lot faster these days.
You can use the REGEXP_REPLACE() function with no additional flags.
select regexp_replace('1. xxxxxx1. xxxx1. ', '1. ', '');
returns:
xxxxxx1. xxxx1.
db<>fiddle here
