On 25/10/2017 4:22 PM, Windt, W.K.F. van der (Fred) wrote:
Did you mean REXX the "lame" horse?:) Only joking! I was reading
about the new "z/OS client web enablement toolkit" which comes with a
REXX API for HTTP requests and a JSON
>
serializer/parserhttps://www.ibm.com/support/knowledgecenter/en/SSLTBW_2.3.0/com.ibm.zos.v2r3.ieac100/ieac1-cwe-json.htm.
It's great that IBM are writing APIs like this to enable languages
with limited capabilities to do modern stuff, but I couldn't help thinking
that's a lot of faffing around calling multiple services just to yank some data
from from a JSON string. It's interesting to compare that approach to other
scripting languages that run just fine on z/OShttp://groovy-lang.org/json.html.
We also stumbled upon a serious flaw in the JSON parser: you cannot parse any
JSON that may contain attributes that can occur on different levels under the
same object: for example:
{"a":1,"A":{"a":2,"b":3}}
If you try to look up the value for "a" in the main object the parser will return either
1 (for the attribute a in the root object) or if the root does not contain attribute a value 2 (for
the attribute in object A). Without any indication which attribute was found. There is now way to
indicate that you want to look for an attribute directly under an object. We reported this to IBM.
They acknowledged that it is not possible to correctly parse these JSON streams but that the parser
was "working as designed" and that we should register an RFE. We did that a few months
ago but have not heard any on that.
OMG! That's a shocker. The API is basically broken as designed! Is your
application CICS/COBOL?
Needless to say that we are not very impressed with these new features.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
