News:

• 2015-11-10: moved to GitHub
• 12.3.2002: RAIN 1.5 have been released.
  Default DSound buffer length increased to 200ms (100ms is short on some computers).
  When no sound device is detected, RAIN doesn't continue silently nor doesn't complain about "Device busy" as it did in Windows, it terminates immediately with exitcode 98. Application loader2 recognizes this code and runs application without RAIN immediately.
  What does it mean? App has more memory when there is no soundcard.
• 27.9.2001: RAIN 1.4c have been released. New high-level C/C++ API allows clients to be written in C/C++. Both C/C++ and Pascal clients may be also written in Windows. See current state of project for list of supported compilers/platforms (new: DJGPP, MinGW, Cygwin, Watcom, FreePascal, TMT Pascal). Btw, C/C++ API is direct port from pascal API, 2 evenings of work; and complete support for Windows clients wasn't more than inserting some ifdef. Servers (version 1.4) remain unchanged. Documentation redesigned.
• 4.7.2001: RAIN servers 1.4 have been released. Multiple clients may be served sequentially by one server started with "-p" parameter (you can run "RAIN.EXE -p BP.EXE" and develop in Borland's IDE with RAIN serving to all programs you run). MP3s with RIFF (WAV) header spread like plague so they may be played now.
• 12.1.2001: RAIN servers 1.3 have been released. Enhanced application loader tries to run apps with rain and then, in case of failure, also without rain. MP3s with ID3 header may be played now. DirectSound device is disabled by default (Why? Dsound seems unable to start playing from fullscreen console, console must be windowed for at least one moment. Switching console to window for one moment is too dirty hack for me, disabling dsound is cleaner. Asking at msdn.microsoft.com was absolutely useles...please help if you are guru.)
• 14.2.2000: RAIN winserver 1.2 has been released. DirectSound device is enabled and set by default... Rain uses "sticky focus", it means that it plays even from background, but stops when foreground application uses DirectSound. Server sets high priority class for self (I expect that computers are strong enough to make some music, so let's make it before other jobs to avoid stream interrupts)... but I also expect you won't play twenty mp3s at once. App.com loader is little bit longer...there was no problems reported, but current version is safer.
• 3.1.2000: RAIN servers 1.1 have been released. Read news in new paragraph Configuring. Small harmless bug in rain.pas fixed (random error report). Windows specific: "8bit cards in windows" bug found and fixed in Midas. Default windows buffer length increased from my crazy 100ms back to 500ms. DirectSound still disabled, because I don't know how to make it sound even in fullscreen and not just in window. DOS specific: Interrupts enabled inside Rain.
• 20.12.1999: RAIN 1.0 has been released.
• 25.11.1999: Bug in NT or Watcom found, infinite waiting for child process fixed (command CMD_TERMINATE must have been added). DOS server uses timer again, but doesn't speed it up, it's free for application as long as they call old vector sometimes, polling is needed again, but only for mp3 playing in DOS. General protection faults seen no more. DirectSound code disabled for poor functionality, wave device stays. Loader app.com fixed, now it has 265 bytes (256 would be better). It seems that everything important works now, expect soon release if no new problems appear during final tests.
• 27.9.1999: Servers and utilities exit codes specified. Watcom wstub.exe bug found, wstub.c different but also buggy, stub example in help different again but also buggy, new stub written, now RAIN.EXE is windows server and RAIN.LE dos server. RAIN.EXE has stub that calls DOS4GW.EXE RAIN.LE. Applications may use timer as they wish, modified Midas uses it no more. Everything (not just mp3) now depends on polling. Problems with taskswitching seems to be only Gravis Windows drivers problems, SoundBlaster plays well. Windows server reports general protection faults sometimes during its shutdown. It was seen only with Gravis...is it Windows drivers problem again?... more time needed before release.
• 24.9.1999: Protocol is set up by environment string rather than by commandline parameter. MP3 volume fixed. Some problems remain to solve... taskswitching can stop sound output..maybe just bad Gravis Windows drivers?
• 22.9.1999: MP3 support added. Standard way of running applications defined, see 'Running applications' section below. Another Midas bug located and fixed, amp bug (also in cubic player) still escaping. Work complete, sources clean, makefiles dirty...no makefiles, it's quite hacking with midas and amp. Not tested under NT yet. Wait for soon release.
• 13.9.1999: One Midas bug located and fixed, nice high-level API added :)
• 9.9.1999: No news yet, all this is fresh enough :)