Double headed GMT4 vectors loose one head from Julia

[joa-quim]

This works fine and plots the double headed vector (set by the -SvB)

echo 1 0 45 2 | gmt plot -R0/3/-1/1 -Jx1 -Bafg -BWSen -N -Gred -W1,blue -SvB4p/18p/7.5p -png lixo

However, replicating it in Julia looses one head and shifts locus

plot([1 0 45 2], R="0/3/-1/1", J="x2", B="afg WSen", G=:red, W="1,blue", S="vB4p/18p/7.5p", show=1)

[PaulWessel]

Hard to see how this can be a C bug. The flag to do dual heads are in the option, not the data file, so should not matter that you give the coordinates via a matrix. If you do the verbose and see the Julia command string passed to GMT, is it missing something or is it actually passing the correct option?

[joa-quim]

When I use strings in options values there is no parsing. We send almost directly the GMT command.

plot([1 0 45 2], R="0/3/-1/1", J="x2", B="afg WSen", G=:red, W="1,blue", S="vB4p/18p/7.5p", show=1, Vd=1)
        psxy  -R0/3/-1/1 -Jx2 -Bafg -BWSen -Gred -W1,blue -SvB4p/18p/7.5p -P -K > C:\TEMP\GMTjl_tmp.ps

[PaulWessel]

What do you think is happening here. The data set is not consulted regarding what is being plotted. The two-headed comes from the -Sv string. Maybe you can step through and see what is different ion the two cases since I am at a loss.

[joa-quim]

I'm at loss too. The problem for debugging is that from julia it became nearly impracticable since each step takes ~20 sec.

[PaulWessel]

OK, here is what happens. GMT defaults for GMT_COMPATIBILITY is 4 and with that the CLI runs as you show. However, the Julia call has GMT_COMPATIBILITY = 6. That is the difference. So what. But we have this strange check in gmt_init.c:

if (!gmt_M_compat_check (GMT, 4) || (strchr (text, '+') || !p->v.parsed_v4)) { /* Check if new syntax before decoding */

This test is true for Julia (since 6 > 4) and it does not check that parsed_v4 is already true... Inside the test there are parsing for modern vector syntax and things go wrong quickly. I suspect this ought to be

if (!p->v.parsed_v4 && (!gmt_M_compat_check (GMT, 4) || (strchr (text, '+'))) { /* Check if new syntax before decoding */

That says "if we did not already detected V4 syntax vector earlier then we can check fir new syntax here. Either we find a + or compatibility mode > 4 and in both cases we expect new syntax."

Seems reasonable (before actually doing the coding)?

[PaulWessel]

My suggestion seems to work. I will make a PR.

[joa-quim]

Yep, tried that and it worked.