+ as it was removed in master

+ iks-server is much more stable than at its beginning
+ supervisor is doing more harm than good to iks-server

Conflicts:
	debian/szarp-iks-server.postrm
	debian/szarp-iks.init
	resources/parstart_TEMPLATE

+ assertion handler throwing exception
+ fixed bug killing iks-server

+ this is really a debug message, making imuxsock drop messages from paramd, so it should have debug priority

Due to a bug in mingw64 libc implementation, it is not
possible to pass NULL as a second argument to wcstod,
as it leads to nullptr dereference...

make sure that boost::remove_all error is ignored

Kick out recently added timegm definition. This defintion was
not only unnecessary (we already have one in mingw_missing.cc)
but also is incorrect. With that new timegm szbase wasn't working
at all on e.g. montly period.

it is again possible to switch period types using time
widget

...is clicked

local2szarp conversion routine was including 0 at the end resulting
std::wstring, which was e.g. messing up boost::filesystem::wpath
operations on such strings

also - cosmetics for utf2szarp for MSW

+ without this a supervisor restart would not relaunch iks-server
+ parstart start/stop works still fine with iks-server

Uzycie opisane w readme.

reksio authored 9 years ago and Aleksy Barcz committed 9 years ago

this fixes a bug where 10 sec blocks for lua optized params
could be created with incorrect start time

reksio authored 9 years ago and Aleksy Barcz committed 9 years ago

due to sz4 refactoring the refresh of drawdefinabe params
for 10sec probes stopped to work correctly

this fixes a bug where 10 sec blocks for lua optized params
could be created with incorrect start time

due to sz4 refactoring the refresh of drawdefinabe params
for 10sec probes stopped to work correctly

- slightly better naming for unit state
- avoids scheduling query when in IDLE state

- remove per param time diffs caching, it is taking too much space
- change heartbeat handing, so that it cannot be starved by huge
  data inflow

this adds simple optimization to meaner4 - if meaner4
cannot keep up with incoming data, it will combine
consecutive messages carrying the same values' into
one write

meaner4 behaved incorrectly if no-data was received as first
datum for a param

filter out not sz4 file in find_latest_path method