A bug in rtklib2.4.3 B34 single point positioning

A problem in the single point positioning part

This is a new problem introduced in this update , There was no such thing before , Here's the picture ：

• If you don't recompile the code, you won't encounter the problems mentioned in this article , because release There will be no problem with this version of the program
• If the code is debug, Then we won't encounter this problem
• If you try to run directly debug Version of the program , Then the probability of this problem
  

A platitude root cause

This issue It is caused by local variables not being initialized , So I want to fix This issue It's also very simple. , Just put these four variables dion,dtrp,vmeas,vion,vtrp Initialization . From an algorithmic point of view , These four variables will not be initialized only in the first iteration , So we initialize it to 0.

static int rescode(int iter, const obsd_t *obs, int n, const double *rs,
...
{
    
    gtime_t time;
    double r,freq,dion,dtrp,vmeas,vion,vtrp,rr[3],pos[3],dtr,e[3],P;
   ...   
        if (iter>0) {
    
            //  The ionosphere is calculated here   Process delay and variance 
        } 
        /* pseudorange residual */
        v[nv]=P-(r+dtr-CLIGHT*dts[i*2]+dion+dtrp); // this is line 306
   ...
}

From a code perspective , There are two aspects of this code that are not very good for ordinary developers （ The code was written by the great God , The great God can write at will ）. These two places are very old-fashioned , Just pay attention to one place , It wouldn't have happened issue.

• Local variables are not initialized , More than local variables , All variables are initialized , This is really a very good habit
• if No, else. There are two possible cases else It doesn't make any sense .
  – if(true),ture It can also be an expression, etc , Whatever the consideration, this situation arises , He showed up anyway
  – else You don't need to do anything .
  In both cases , Write not write else There is no difference in program execution . But there is a big difference between avoiding mistakes and reading code . Because as long as you write else It proves that the author of the code considered else This situation , This is a deliberate act . So at this time else It is suggested to write in the following form ,

if(...) {
    
//...
} else {
    
    NULL; // REMINDER!:do nothing on purpose
}

Or the following form is acceptable ,

if(...) {
    
//...
} else {
    } // REMINDER!:do nothing on purpose

Why is it so modified

So why did the author make the following changes ？ It's simple , Let's take a look at the code , This if I did the following four things , These are all quantities related to the position of the receiver . Although this position does not need to be very accurate , Rough coordinates are enough . But we know it's pvt When solving , The initial position is unknown , There is no point in calculating these quantities here , And the result of calculation is wrong , It has side effects , It might as well not be . This change not only saves CPU, My guess is that it will also improve iteration efficiency .

• Elevation mask
• SNR mask
• Ionospheric delay calculation
• Tropospheric delay calculation

if (iter>0) {
    
    /* test elevation mask */
            if (satazel(pos,e,azel+i*2)<opt->elmin) continue;
            
            /* test SNR mask */
            if (!snrmask(obs+i,azel+i*2,opt)) continue;
            
            /* ionospheric correction */
            if (!ionocorr(time,nav,sat,pos,azel+i*2,opt->ionoopt,&dion,&vion)) {
    
                continue;
            }
            if ((freq=sat2freq(sat,obs[i].code[0],nav))==0.0) continue;
            dion*=SQR(FREQ1/freq);
            vion*=SQR(FREQ1/freq);
            
            /* tropospheric correction */
            if (!tropcorr(time,nav,pos,azel+i*2,opt->tropopt,&dtrp,&vtrp)) {
    
                continue;
            }
        } 

So change

therefore ,fix The method will be natural , Maybe it's changed like this ,

    dion = 0; dtrp = 0; vion = 0; vtrp = 0;
    
    for (i=*ns=0;i<n&&i<MAXOBS;i++) {
    

Or so ,

if (iter>0) {
    
            //  The ionosphere is calculated here   Process delay and variance 
} else {
    
    dion = 0; dtrp = 0; vion = 0; vtrp = 0;
}