Re: [Ortcan-devel] bug fixed

[Pavel P. <pi...@cm...>]

On Thursday 10 May 2012 16:12:22 Ivan Fridrich wrote:
> I fixed EDS file saving bug in QCANalyser.
>
> I'am going to integrate EDS check utility into QCANalyser.

Thanks for your work. I have tested the most recent
QCANalyser version and it works much better.

I have run next tests sequence

I have load "nascan.eds" file.
Then I have saved it under "nascan1.eds" - as exected, order
of many subfields is alterred according to order of storage
in the tree.

I have read it again and stored under name "nascan2.eds".
We should not observe changes when do read and writes
by same tool. This was observed behavior except
lost of already reduced initial comments lines

--- nascan1.eds 2012-05-10 22:38:42.297078000 +0200
+++ nascan2.eds 2012-05-10 22:39:42.942077999 +0200
@@ -1,7 +1,3 @@
-[Commments]
-Line1=|---------------------------------------------------
-Lines=8
-

The original file is incorrect probably, because it contains
multiple

  Line1=............

lines. But even last one has been lost.

Even that lines order in each section can be alterred,
we can observe, if some lines are lost or changed
during EDS parse and save by QCANalyser

  sort nascan1.eds >nascan4.eds
  sort nascan.eds | grep -v '^;' >nascan3.eds

diff -u -B -b nascan3.eds nascan4.eds | less

There are some differences which should be resolved

-[100A]
-[100B]
-[100C]
-[100D]
-[100E]
-[100F]
+[100a]
+[100b]
+[100c]
+[100d]
+[100e]
+[100f]
...
-[2900sub6A]
-[2900sub6B]
-[2900sub6C]
-[2900sub6D]
-[2900sub6E]
-[2900sub6F]
+[2900sub6a]
+[2900sub6b]
+[2900sub6c]
+[2900sub6d]
+[2900sub6e]
+[2900sub6f]

Original file uses uppercase letters in hexadecimal representation.
It should be checked against specification what is the correct or suggested
representation. This problem is not so critical.

But switching of the whole stream to hexadecimal base has unfortunate
consequences

[OptionalObjects]
...
-10=0x100C
+a=100c
-11=0x100D
+b=100d
...
[ManufacturerObjects]
...
-10=0x2900
+a=2900
-11=0x3000
+b=3000

The default base for values should be decimal or 0x has to be preceded. Key names (if numeric)
should be expressed as decimals too.

Next blank character removal is OK

-ParameterName= 27. 1byte input
+ParameterName=27. 1byte input

But next changes are bugs

 [OptionalObjects]
-SupportedObjects=42
+SupportedObjects=2a

I think, that only indexes and subindexes should be expressed
in hexadecimal and without 0x prefix.
The rest of the data should be saved in decimal only.
Or 0x prefix could be used but only for values which has been
in hexadecimal in the input file.


But generally the EDS editor implementation is significant
step forward.

Best wishes,

               Pavel