Menu
▾
▴
[netrexx-pipelines] SPECs - Working
|
From: Jeff H. <Je...@Je...> - 2020-03-28 20:53:32
|
<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
</head>
<body>
<p>I have just submitted an updated SPECs stage and tests.</p>
<p>There are 84 separate tests in 8 files. There are a couple from
the IBM Pipeline book that we still fail.</p>
<p>Here is the current documentation:</p>
<p><tt>/** spec<br>
+-STOP--ALLEOF----+ (3)
v----------------------------------------+<br>
>>--SPECs--+-----------------+------+--+-| Group
|-------------------------+-+-----------><<br>
+-STOP-+-ANYEOF-+-+ (3)
+-READ------------------------------+ (5)<br>
+-n------+ (3)
+-READSTOP--------------------------+<br>
+-WRITE-----------------------------+<br>
+-SELECT-+-streamnum-+--------------+<br>
|
+-streamid--+ | (3)<br>
|
+-FIRST-----+ |<br>
|
+-SECOND----+ |<br>
+-PAD--+-char----+------------------+<br>
|
+-hexchar-+ |<br>
|
+-BLANK---+ |<br>
|
+-SPACE---+ |<br>
+-+-WORDSEParator---+--+--char---+--+ (3)<br>
+-WS------------- +
+-hexchar-+ (3)<br>
+-FIELDSEparator- +
+-BLANK---+ (3)<br>
+-FS------------- +
+-SPACE---+ (3)<br>
<br>
Group:<br>
|--| Input |--| Conversion |--| Output |--| Alignment |--|<br>
<br>
Input:<br>
|--+-Words-(1)-wnumberrange----------------------+-------|<br>
+-Fields-(1)-fnumberrange---------------------+ (3)<br>
+-cnumberrange--------------------------------+<br>
+-/string/------------------------------------+<br>
+-Xhexstring----------------------------------+<br>
+-Hhexstring----------------------------------+<br>
+-Bbinstring----------------------------------+<br>
| +-FROM--1-------+ +-BY--1-----+ |<br>
+-+-RECNO--+-+---------------+-+-----------+--+<br>
| +-NUMBER-+ +-FROM--fromnum-+ +-BY--bynum-+ |<br>
+-TODclock------------------------------------+<br>
<br>
Conversion:<br>
|-+-------+--+--------------------+------|<br>
+-STRIP-+ +-B2C----------------+<br>
+-B2D----------------+ (4)<br>
+-B2X----------------+ (4)<br>
+-C2B----------------+<br>
+-C2D----------------+<br>
+-C2F----------------+ (3)<br>
+-C2I----------------+ (3)<br>
+-C2P-+------------+-+ (3)<br>
| +-(2)(scale)-+ | (3)<br>
+-C2V----------------+ (3)<br>
+-C2X----------------+<br>
+-D2C----------------+<br>
+-D2X----------------+ (4)<br>
+-F2C----------------+ (3)<br>
+-I2C----------------+ (3)<br>
+-P2C-+------------+-+ (3)<br>
| +-(2)(scale)-+ | (3)<br>
+-V2C----------------+ (3)<br>
+-X2B----------------+ (4)<br>
+-X2C----------------+<br>
+-X2D----------------+ (4)<br>
+-f2t----------------+<br>
+-LOWER--------------+ (4)<br>
+-UPPER--------------+ (4)<br>
+-STRING-------------+ (4)<br>
<br>
Output:<br>
|--+-Next-+-------+-------+--|<br>
| +-(2).n-+ |<br>
+-NEXTWord-+-+-------+-+<br>
+-NWord----+ +-(2).n-+ |<br>
+-collumnrange---------+<br>
<br>
Alignment:<br>
|--+--------+--|<br>
+-Left---+<br>
+-Center-+<br>
+-Centre-+<br>
+-Right--+<br>
<br>
Ranges (cnumberrange, fnumberrange (3), wnumberrange):<br>
|--+-snumber-+--(2)--+-------------------------+--|<br>
+---*-----+ +--.-----(2)----number----+<br>
+- - -+--(2)--+-snumber-+-+<br>
+--;--+ +---*-----+<br>
<br>
Notes:<br>
(1) Blanks are optional in this position.<br>
(2) Blanks are not allowed here.<br>
(3) CMS only. Not yet implemented in njpipes<br>
(4) Njpipes only. Not yet implemented in CMS<br>
(5) READ is giving the same output as READSTOP when the streams
are different length.<br>
[6] This senses if it is the first stage, but comment stages
will fool it into not<br>
producing any output.<br>
<br>
**/</tt></p>
<p>The one bug left that I'm aware of, is if there is some data
generating group, such as a delimited string, record number, or
TODclock, then at the end of file, an extra output record of just
that generated data is put out.<br>
<tt></tt></p>
<p>The only parts of the newest IBM version that I have included
here are<tt> FIRST </tt>and <tt>SECOND. </tt> This newest
version also includes a mini-language for doing all sorts of
things in the conversion phase. This would be very difficult to
include here. I suggest that instead we add a way to incorporate
either a sub-pipeline (apparently difficult) or user NetRexx code
(fairly easy, but potentially dangerous). I have a bare-bones
working test of this later: it has, in the conversion section,
user_code in parentheses: <tt>specs /xyz/ (data =
data.changestr('y','A')) 1</tt> does put out <tt>xAz</tt>.</p>
<p>Is such a facility worthwhile at all? Is it worth putting some
effort into implementing it? I'd like to limit it's reach to only
the variables with the suggested names of <tt>data</tt> (input
and output -- or should these be different?) and <tt>user</tt> a
"persistent" indexed variable available to store things across
groups in a call and across records in a pipe. At the moment, I
don't know how to do this isolation.</p>
<p>The other new feature that looks worthwhile implementing is EOF
handling. I have not looked at how hard this would be. (And at
the moment, I have had enough time with SPECs!)</p>
<p>The whole thing <u>should</u> be redone at some time using the
IRange and Dstring classes for parsing the input.</p>
<p>Jeff<br>
</p>
</body>
</html>
|
