Re: [mkgmap-dev] methods to write signed / unsigned integers

16 Mar 2018

      Hi Ticker,

yes, I think display tool needs more work. You commented a
I think the method names in Displayer.java should be changed so that they are similar to the corresponding reader methods.
Example:
        public int int3Value(String text)
would be
        public int int3uValue(String text)

I see some places where you replaced
                item.setBytes(reader.getChar());
by
                item.setBytes((char)reader.get2u());
I think the cast to char should be removed.

I also see code like this:
                                c = item.setBytes(reader.get2u()) & 0xffff;
I think the & 0xffff is obsolete and confusing.

I am pretty sure that some lines in MdrDisplay are wrong (not because of your patch).
I think all places where get3s() is used need a closer look. We might have to change them to get3u().
For example:
 int record = reader.get3s();
off = item.setBytes3(reader.get3s());
Record numbers and offsets are normally not negative. I'll have a closer look today.

I also agree that this snippet from NodConvert looks wrong:
		} else if (restrbytes == 3) {
			size = reader.get2u() & 0xffffff;  // %%% think mistake

I've not used NodConvert for a long time, it might well be behind the last findings reg. NOD.

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces@lists.mkgmap.org.uk> im Auftrag von Ticker Berkin <ticker@jagIT.co.uk>
Gesendet: Donnerstag, 15. März 2018 15:21:25
An: mkgmap-dev@lists.mkgmap.org.uk
Betreff: Re: [mkgmap-dev] methods to write signed / unsigned integers

Hi Gerd

I've attached the Display patch - this was a quick hack global edit and
could be improved by using the relevant signed/unsigned get methods
instead of masking...  I'm happy to do this - let me know.

Yes to DEM changes. Also the nearby asserts can go now because put2s()
will check the range.

  // don't think needed:
could be changed to
  // not needed at moment:
but I think it is worth leaving these methods (getNs()/putNs())
commented in the ImgFile interface source.

Ticker

On Thu, 2018-03-15 at 10:12 +0000, Gerd Petermann wrote:
...
Hi Ticker,
sorry, did not find the time to look at it until now.
At least for DEM this would not be correct, some values should be
written with the signed put versions.
I've attached a corrected version of the patch.
Besides that I'd prefer not to see comments like
// don't think needed:
For further testing I'd need the patch for display tool as well.
Gerd