[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: RFS: haskell-tz and haskell-tzdata



On Tue, 21 Jun 2022, Robert Greener wrote:

On Tue, 2022-06-21 at 10:14 +0300, Ilias Tsitsimpis wrote:
On Mon, Jun 20, 2022 at 10:47PM, Robert Greener wrote:
That wasn't what I found, maybe it has changed? It appears to
provide a
-data package that contains the tzdata files. I've patched this to
remove the data package and to just load the files from the
standard
location. I've copied the patch below -- what do you think?

I see. So you patched the 'timeZonePathFromDB' function.
Unfortunately,
this function is used only for testing:

$ grep -rn timeZonePathFromDB
Data/Time/Zones/Files.hs:10:  timeZonePathFromDB,
Data/Time/Zones/Files.hs:19:timeZonePathFromDB :: String -> IO
FilePath
Data/Time/Zones/Files.hs:20:timeZonePathFromDB tzName =
tests/testDB.hs:15:  pathBp <- timeZonePathFromDB "Europe/Budapest"
tests/testDB.hs:23:  pathEtc <- timeZonePathFromDB "Etc/GMT+1"

i.e., they use this function in order to read the tzdata files and
compare them with what data they have hardcoded.

Take a look at Data/Time/Zones/DB.hs. You will see that this file
contains all of the tzdata as hardcoded strings. See also this script
https://github.com/ysangkok/haskell-tzdata/blob/master/build-tzdata.sh
that they use in order to build this file, before uploading to
hackage.


Ah, thanks. I'll have a think about what to do here...

In the mean time, I'll patch arbtt so that the current version in
debian can use the latest version of aeson.

Hi again Robert,

Did you end up having any further thoughts on this?

It seems we may have more packages wanting to depend on tz, so we may need to come up with a solution.

Scott

Reply to: