IO2 Updates History

"SQ" applies to both, SMSQ and SMSQ/E, w.r.t. any one of the many versions.

SIGNIFICANT CHANGE due to restructured PEX code from PEX20 on.
Thus, never use a version of PEX prior to PEX20 with IO2 versions after July 1996!

• NOTE on loading IO2:
  • The code, at least DVS_BYT if single blocks loading, should be loaded AFTER the Pointer I.F.
• yet uncorrected:
  
  • W_CLIP% scanning a "D" as the 1st character can return a "[" or even nothing. Sub-/Superscript recognition is very uncertain.
  • SBAS detection depends on the actual job structure and might lead to problems when executing compiled basic in SMSomething environments.
• system related errors, once more (SMSQ, SGC with rom 2.49, etc.):
  • Some more recent versions again bear undocumented changes and mistakes (as SGC/2.49 not formatting reliably; SMSQ/E deliberately, apparently, loosing track of slave blocks, returning 'bad medium', etc.) which to first reporting and, after receiving no action at all or just denial of the bug concerned, circumventing from within the IO2 codes over and over again I am tired.
    Thus there will be no more IO2 adaptations to those, mostly SMSQ/E specific, "peculiarities", nor will there be any response to fancy "features", whatsoever, which are not clearly documented!

• updates, bug fixes, SMSQ related adaptations:
  
• 01/09/95
  • BEEPING new, to work while TUNE is waiting (pause "note")
  • FIND err.xp (-18) if length limit exceeded
  • LINUM also reading Qlib-compiled running line numbers
  • parameters IO1 parameter evaluation made faster
  • W_PIXCOL etc. WD1 graphic commands modified to working with JM systems
• 04/09/95
  • CH_PILE 3rd -ve parm for own primary window
• 09/09/95
  • CMD dummy channel -1 to resetting to system standard
  • ST_EXECB stack bounds adjustment on return left to user
  • STAK new: trap#3 opcode $41 to execute a call to ri.execb
• 18/09/95
  • C_N big-/litle endian order & no. of valid bits selectable
  • W_XMIN W_YMIN new
  • W_YSCN bug fixed
  • POKE.. all executing in supervisor mode (SMSQ feasiblity)
• 22/09/95
  • GET$ memory allocation bug fixed
  • REPORT$ SMSQ extended set of messages inclusive
• 23/09/95
  • LANGUAGE$ SMSQ system-returns remain unchanged
• 24/09/95
  • VERS genaralized VER$ function, to circumventing a bug in JS
  • MODE separately initiated by MDE_INI (since SMSQ-2.68 bp.init works as expected).
  • WMON functionality of internal call to MODE restored in case of re-definition of the CON-handler; needs PEX loaded.
• 26/09/95
  • character founts in DVS hi-/lo-script modified and extended
  • mv.vlet memory allocation bug fixed
• 30/9/95
  • all IO2 blocks memory allocation bv.rip re-/storing bugs fixed
  • ST_EXECB circumvents a JS and MGx roms bug (at adddress $4530), calls to ATAN etc. are always safe now, at the penalty of slightly slower executing. Consequently run-time efficiency of ST_ECECB exceeds that of SBasic only with fairly long calculation sequences.
  • W_CLIP$ W_CLIP% reading non standard height increment ok, and with JM
  • W_PRINT SQ: parameters memory properly linked back to system
• 10/10/95
  • IO_FLINE basic-memory and string length bug fixed
  • W_LINEL non-standard screen width bug fixed.
• 16/10/95
  • GET$ direct sector read enabled.
• 10/11/95
  • ASO automatic display dis-/enabling left to be initiated explicitely by AS_LNK.
• 18/11/95
  • ASO another attempt to safely initiate the timer interrupts on atari-tt's by adding AS_GO to explicitely initiate the procedure.
  • QL_ROM modified to comply with atari's memory-"protection".
• 19/11/95
  • AS_GO, AS_LNK once more adjusted to the SMSQs,
    ASO actions left to be initiated by AS_GO and AS_LNK, one after the other.
    handling of pause-"notes" adjusted to SMSQ, again...
  • SOUND$ new function to reding the currently playing "note".
• 24/11/95
  • SER_IS modified to alternatively work on "par" devices,
  • SERNAM$ new, to return the name currently used by SER_IS,
  • SERNAM new, to supply new ser/par-devicenames to SER_IS
  • dvs io1 buffer-pipe definition extended to take that name.
  • TUNE pause-flag storage bug fixed at <tun_poff>
  • aso finally, I had to give up:
    automatic screen switching dispensed with for any other system than those of the black QL and the e.xx Atari-QL.
• 22/3/96
  • io2 SQ-pre-2.68 version bugs processing removed.
    "SBAS" detection relating to the calling (basic-)job.
  • DMODE screen disabling bit untouched in read only mode.
    execution by mv.mrun with unchanged con-handler.
  • SERNAM$ additional parameter as with ser_is.
  • BYE new, synonym to QUIT.
• 14/4/96
  • dvs/version BX_..-calls show a different version specification.
  • aso A4 saved at calls to mt.dmode (JS-bug, SQ-"feature");
    automatic unlinking dummy device on owner-job removal; re AS_TAB
    nn=FOPEN("ASO") or type 0 io.open-call returns address of aso database ( = AS_TAB - 8 ).
  • EDLIN & EDLIN$ initially timing out if 4th parm passed and +ve.
  • CH_CHS[#]chan[,hash[,jobno|jobid]
    new: displaying channel-id's & -names to [#]chan.
  • IO_FLINE string length determining bug fixed.
• 17/6/96
  • dvs mayor work on MEM-device, virtual addressing prepared.
  • GETHEAD$ padded with <nul>-bytes up to the length of 64.
  • LINKUP (and others) memory management fixed.
  • io2 calls to fs.load/save modified to io.fstrg/sstrg to error free operating with non-standard formatted disks (where some fs..-calls can produce very bad mistakes).
• 5/7/96
  • CH_CHAN$ ser-type & -parameter recognition completed, also returning the SMSQ/E specific settings.
  • CH_CHS timeout 2,5s (for SER-/PAR-output) intead of 40ms, additionally made "break"-sensitive (CTR/SPACE).
  • BIGL_E removed.
• 17/7/96
  • BASICV_byt added (freeware);
    much like the original procedure supplied with FIO, or VGET, additionally fetching the 1st (index 0) item of dimensioned variables.
  • DVS_byt vectored routines adjusted to conform PEX20+ structure.
• 31/10/96
  • CH_CHS stack error if IO2-buffer-PIPE was not found corrected.
• 4/11/96
  • SER_SEND modifying a SER channel to not receiving, speeding up the system if the other port bi-directionally in use.
  • VERR$ VNUL$ discarded (was useless), new additional error handling; mainly to act on procedure call errors, as any function return values will be lost:
  • THROW executes VGET as a procedure and only if ERNUM =/= 0; it returns with no error, leaving ERNUM unchanged.
  • CATCH executes VGET as a procedure, setting ERNUM to the usual error code, itself returning always error free. Thus
    CATCH "hex$","abcd",19:THROW"print",#3,report$
    will send an error message to #3 and proceed error free.
    NOTE: 'CATCH_F'/'CATCH_P' of the "fio" group are different!
• 11/11/96
  • SOUND$ and related keywords might not work with the Super-GC (rom 2.49!) and most certainly will not work with QPC.
• 27/11/96
  • CH_QKEYQ finds the channel under the pointer sprite, if active.
• 1/12/96
  • dvs version changed to DD22 due to new charcter set
  • EMUFOUNTS_BIN new, may NOT be loaded to old versions of dvs, while loading old charctersets to the new version does no harm, but might corrupt the display.
  • FT_LAT1 added, for the new LATIN-1 (ISO8859) charcterset. "founts_dat" updated accordingly.
• 19/12/96
  • LINUM extended to taking a job-descriptor parameter for returning a QLib-compiled jobs current line number.
  • IS_PNTR etc; PIF-recognition completed, bug fixed.
• 3/1/1997
  • QXL_XTD error corrected
  • CATCH_.. operating w. interrupts locked to prevent erroneous addresses when called while jobs changing (fast).
• 7/2/1997
  • ATARI new: 1 in Mega-ST, STE, TT, Falcon, Stacy; -1 as above, with SMSQ; 0 otherwise (QL, ST, QPC)
  • AS_GO etc. (automatic) display blanking functional again, with Atari-ST and e-software and SMSQ/E (2.79 tested).
  • AS_TIME optionally with an integer number parameter will set the rate of, and return the previous setting in ticks per second, for the screen disabling, TUNE, and TICK timers.
  • COMP COMP_O videoram addressing simplified, sending file data bug fixed.
• 11/2/1997
  • CLOSE_B% closed channels *Basic flag bug fixed.
  • CATCH made functional.
  • WORDS as EXTRAS, tabulating to the full width of display, taking the WIDTH adjustment of SBasic channels, or the display windows width. Channel parameter as GET$. <esc> for compiled and interpreted jobs to <break>.
  • CATCH_OFF address error trapping improved.
• 9/3/97
  • MT_TRA prepared to working also with pre-JS systems.
  • mtra_tab char 127 (SH/ESC) translated to "(C)".
  • CH_CHAN$ EDxx recognition modified,
    SER/STX/SRX of "simser" added (SMSQ/E-ish specification)
  • BPUT modified to sending uncounted characterstrings and byte codes, plus extended channels specfication as GET$.
• 28/4/97
  • DEVUSE$(dev) Fu to return current devicename usage (QL-Today 6/97).
  • AS_GO etc. (automatic) display blanking functional again, with QXL and SMSQ 2.76 or SMSQ/E (2.79 & 2.81 tested). This occupies some extra memory space, depending on full screen size (bytes of line length * no. of lines).
  • AS_ON AS_OFF Initial state is AS_OFF; AS_ON used to activate the screen blanking - before regular use test whether the particular system permits this mode of screen blanking!
  • POKE_N bug(s) fixed.
  • MEMV_BYT 1st attempt to adjust to SVGA screen mode (QXL/AURORA) -> ok with QXL, no result with AURORA
  • PACODE bug fixed.
• 12/12/97
  • LINUM(job) with reference to interpreted SBasic jobs implemented
  • CH_CHAN$ STX/SRX and EDI-channels fetching
• xx/5/98
  • CH_CHS at least one blank space inserted between data blocks
  • io2 error with "false" channel tags fixed: -ve tag numbers are permitted (re. MGG)
  • dvs 2nd attempt to using virtual memory, being invoked if requested amount of heapmem not available.
    new traps:
    sd.fount A1=>device A2=>filename, to preparing v.m.
    S.Basic: CHAR_USE#mem-channel,"device","filename"
    fs.date always current
    fs.mdinf fs.trunc fs.xinf standard
    not much tested, use with care, if ever..
  • MEMV_BYT another attempt to make it displaying with 'big' screens, ok now with the UQLX-emulator.
• xx/6/98
  • CH_FIND leading 2 number parameters may be omitted:
    qch = CH_FIND( [[ job-id,] topmost-chan-no,] name$ [,joker] )
• 9.7.98 : io1
  
  • IO_FLINE as IO_FSTRG
    : wd1
    
  • W_RPTR string$=W_RPTR([#chan,]string$)
    read 32 bytes co-ordinates & pointer record, X/Y-posn in 1st 4 bytes of strg, in and out
• 24.7. : fio
  
  • REBOOT experimental:
    flush & close all channels, and re-start the system (software-reset).
• 10.10 : wd1
  
• modified LINE commands, also taking array parameters at any posn in the parameter list, aligned to x/y-pairs:
  
  • LPLOT LINE of consecutive points;
    Procedure made to displaying the "hershey" founts.
    LINE [#chan], p1: LINE TO p2 TO p3 TO ...
    is LPLOT [#chan],x1,y1,x2,y2,x3,y3, ...
    A new line beginning after x(n)=-50 and y(n)=0.
    
  • LPLOT_P LINEs of pairs of points
    LINE p1 TO p2; p3 TO p4; ...
    is LPLOT_P x1,y1,x2,y2,x,y3,x4,y4, ...
    
• 24.3.99 : fio
  
  • NFSUSE$ (with TK2 2.12+, or GC, or smsq 2.76)
    ret string with the NFS_USE settings
    parm 0..7 : a particular NFS drive;
    parm > 7 : the nfs devicename;
    parm -ve/none : all settings, "," separated.
    example:
    FOR f=0TO 7:PRINT NFSUSE$(8);f+1,NFSUSE$(f)
    
  • DEVUSE$
    modified to return DEV-useage when called w/o parm$
• 22.11.1999 : ep0
  • FIND
    S.Basic memory related bug fixed.
    Searching long files or memory areae not locking up the system.
• : sv0, dvs
  • WMON
    modified call removed

.hpr'98