Based on your pictures let me assume the bitOrder is mostSignificantBitFirst.
The way you construct a binary int from the 10 bits is not the way you are showing it. First you create the contiguous set of 10 bits. This is done by appending all the bits together according to the bitOrder. So the 10 bits are 1111000 000 The first 7 of those are from the first byte. The last 3 of those came from the 2nd byte. Now you divide those up into 8 bit bytes. Since bit order is mostSignificantBitFirst, you do this starting from the left. 11110000 00 Now pad the 2nd byte so it is a full byte. Again this is done based on the bitOrder, so those 2 bits are shifted right by 6 bits. 11110000 xxxxxx00 The "x" are zeros, but I'm using "x" just to illustrate the shift. This is the confusing and subtle step due to bitOrder MSBF. When using bitOrder MSBF, a chunk of bits will come from the most-significant bits of the next byte, but if that's the last part of the number, the place-value of those bits is wrong as far as what they contribute numerically to a number. They have to be shifted to the other end of the byte to contribute the proper numeric value. Having done this shift, it is only now that byte order comes into play. So far, all we were doing was constructing the proper bytes from the bits in the data. Now, treat this as a 2-byte littleEndian integer. So let's write the bits in normal left-to-right place-value order with most significant on the left: xxxxxx00 11110000 So the 10 bit number is 0011110000 or 0x0F0 which is 240. -mike beckerle Tresys ________________________________ From: Costello, Roger L. <[email protected]> Sent: Tuesday, February 19, 2019 6:15:44 PM To: [email protected] Subject: Two bytes ... bitOrder="mostSignificantBitFirst" ... byteOrder="littleEndian" ... I don't understand Daffodil's output Hi Folks, Here’s my input displayed in a hex editor: [cid:[email protected]] So the input is hex 78 01 Here is the equivalent binary: [cid:[email protected]] My DFDL schema declares: byteOrder="littleEndian" bitOrder="mostSignficantBitFirst" My schema declares this sequence of elements: element One, length 1 bit element Ten, length 10 bits element Five, length 5 bits I ran the input plus schema through Daffodil and this is the XML that was generated: <Test> <One>0</One> <Ten>240</Ten> <Five>1</Five> </Test> Please help me understand that output. Let’s work through it. The schema first declares the One element. So, Daffodil consumes the first bit of the first byte: [cid:[email protected]] Is that correct thus far? Next, the Ten element wants 10 bits. There are only 7 remaining bits in the first byte, so 3 bits are consumed from the second byte. Which 3 bits? The 3 most significant bits. Since byteOrder is littleEndian, those 3 bits are more significant than the 7 bits; therefore, the 3 bits go to the front of the 7 bits: [cid:[email protected]] The result after prepending the 3 bits onto the 7 bits is this: 0001111000 That binary corresponds to decimal 120 But Daffodil says the value of Ten is 240. Why? Where did I go wrong? /Roger
