Inverse of posix time

Hello,

O was wondering if there is a function in matlab that does the opposite as posix time. Indeed, I have a UNIX time (1443002431.224) and I want to convert it into a date with hour, minute and second.

How can I do it ? Does it exist a function to do this ?

Thank you very much for your help !!

NoNo

Answer 1

[Guillaume]

   d = datetime(1443002431.224, 'ConvertFrom', 'posixtime')

[NoNo]

Thank you very much ! Very easy answer but I couldn't find it !!

NoNO

[Peter Perkins]

NoNo, if the .224 part is known to be exact (e.g. you read it from text), you've already introduced some round-off in expressing 1443002431.224 as a floating point value. You may not care at all, but if you do, you might consider

>> datetime(round(1000*1443002431.224),'ConvertFrom','epochtime',"TicksPerSecond",1000,'Format','dd-MMM-yyyy HH:mm:ss.SSS')
ans = 
  datetime
   23-Sep-2015 10:00:31.224

As I said, that may be overkil for your purposes.

[Haris K.]

@Peter Perkins if you format the output of the resulted datetime as 'dd-MMM-yyyy HH:mm:ss.SSS' both:

unix_t = 1443002431.224;
datetime(unix_t, 'ConvertFrom', 'posixtime')
datetime(round(1000*unix_t),'ConvertFrom','epochtime',"TicksPerSecond",1000)

return the same output, i.e. 23-Sep-2015 10:00:31.224. So what is the round-off introduced?

[Peter Perkins]

Well, I misspoke, sort of.

The round-off is introduced in unix_t = 1443002431.224. That is not a value that is representable exactly in floating point. However, because of the extremely smart people who created IEEE754 etc. way back when, it turns out that 1443002431.224*1000 is exactly 1443002431224. And that multiplication happens inside datetime, without you needing to do it as in my code. That's what saves this.

I think that should be true for anything specified to milliseconds.

Answer 2

[Emre Güngör]

You need 2018 version

[Steven Lord]

Not true. datetime was introduced in release R2014b and the documentation for that function in that release lists the 'ConvertFrom', 'posixtime' syntax. When I run Guillaume's code in release R2014b I receive the same answer as if I created a datetime for January 1st, 1970 and added that many seconds to it. Both those answers matched the results I received for those same computations in release R2018b and the result I received using an online UNIX epoch converter.

Answer 3

[Anh Anh]

https://www.unixtimestamp.com/index.php

[Rik]

(rescued from spam filter)

Even though this isn't spam, it isn't a real answer either. There are Matlab tools available that do this, so it is not necessary to use an external website to do this conversion.