Fix compiler warnings, part 2

[nilason]

Fixes -Wformat compiler warnings.

One important culprit to these warnings was caused by bad detection of off_t print format with PRI_OFF_T.
Suggested solution works for me. I'm curious if it bears in general.

Second part addressing #1247.

Update:

The PRI_OFF_T problem seems to be Mac-only, updated accordingly.

Modules / code parts directly affected:

• db/drivers/ogr
• lib/vector/Vlib
• raster/r.compress
• raster/r.info
• raster/t.terraflow
• raster/r.viewshed
• raster3d/r3.out.vtk
• raster3d/r3.stats
• tools/timer
• vector/v.in.ascii
• vector/v.lrs/v.lrs.create

[nilason]

The CIs failing because of intmax_t, will see to that ASAP.

[metzm]

The CIs failing because of intmax_t, will see to that ASAP.

Use grass_int64 instead of intmax_t.

[nilason]

The CIs failing because of intmax_t, will see to that ASAP.

Use grass_int64 instead of intmax_t.

I restored grass_int64, but as printf() doesn't recognise that type, I use PRId64 macro for the format. Most platforms (all on CI) seem to support this.

[metzm]

I restored grass_int64, but as printf() doesn't recognise that type, I use PRId64 macro for the format. Most platforms (all on CI) seem to support this.

grass_int64 is not a fixed type, but a platform-specific typedef ... grass_int64 for some available 64 bit integer type. If PRId64 works across platforms, great! If need be, we can define matching macros in gis.h.

The aim to eliminate compiler warnings is great, but bear in mind that some compiler warnings are nonsense, requiring case-by-case judgement.

[nilason]

I restored grass_int64, but as printf() doesn't recognise that type, I use PRId64 macro for the format. Most platforms (all on CI) seem to support this.

grass_int64 is not a fixed type, but a platform-specific typedef ... grass_int64 for some available 64 bit integer type. If PRId64 works across platforms, great! If need be, we can define matching macros in gis.h.

The aim to eliminate compiler warnings is great, but bear in mind that some compiler warnings are nonsense, requiring case-by-case judgement.

PRId64 is declared in <inttypes.h> and therefore dependent on C99 support. Independent of outcome regarding statement on standard Cxx usage it may be implemented in "gis.h" as you suggested. Maybe like G_PRId64? That way it (inttypes.h) need not be included specifically when needed. If you agree, I will put up a suggestion for this.

[nilason]

The aim to eliminate compiler warnings is great, but bear in mind that some compiler warnings are nonsense, requiring case-by-case judgement.

I'm aware many compilation warnings are just annoyances, but if there are acceptable ways of getting rid of them, all the better.

[metzm]

I'm aware many compilation warnings are just annoyances, but if there are acceptable ways of getting rid of them, all the better.

I agree, and try to help to sieve through these compiler warnings. For some of these warnings, I just don't know how to get rid off them without breaking compatibility.

[nilason]

Updated according to your initial review.

The question on PRId64 and <inttypes.h> remains to be cleared. No need to force this, just wanted to fix the other issues.

[metzm]

The question on PRId64 and <inttypes.h> remains to be cleared. No need to force this, just wanted to fix the other issues.

Great, please merge! Let's leave the question on PRId64 and `<inttypes.h> for another PR until there is agreement on the dev ml.

[nilason]

Let's leave the question on PRId64 and `<inttypes.h> for another PR until there is agreement on the dev ml.

Lifted out and moved this part to #1316.

[nilason]

Ready, will merge on approval.

[metzm]

Looks good now.

[nilason]

Thanks a lot. Cherry picked and pushed manually.