| <<<Back 1 day (to 2018/07/22) | 20180723 |
sebras | am I right in understanding that in a non-debug build there are only one level of messages output by gs and that is errors? | 16:42.26 |
| the concept of warning/info/debug/trace/whatever does not apply..? | 16:42.43 |
| but for debug-builds if_debug3m() may be used and then the first argument is some kind of debug level. | 16:43.19 |
chrisl | We do emit warnings.... | 16:44.14 |
sebras | chrisl: but encoders/decoders do not? | 16:44.40 |
chrisl | Not third part ones, no | 16:45.00 |
sebras | chrisl: is jbig2dec a third part one? | 16:45.13 |
chrisl | Sort of, yes | 16:45.41 |
| Although we own it, it's built/treated as third party - much like mupdf does | 16:46.21 |
sebras | chrisl: ok, so then warnings for that should only be printed in debug mode and that requires me to enable messages of class 'w' some how? is this how it is supposed to work? | 16:46.22 |
chrisl | sebras: Do you mean warning, or do you mean debug info? | 16:46.52 |
sebras | chrisl: jbig2dec prints warnings when a file is out of spec, but it able to cope with it. the output might not be the expected one. | 16:47.35 |
| chrisl: in my opinion this is a warning that ought to be printed on the output, and a normal output is also generated. | 16:48.08 |
chrisl | Okay, I would just protect that with a pre-processor define - DEBUG probably | 16:48.39 |
| I don't believe jbig2dec has a nice way to pass error/warning/info messages back to the caller | 16:49.18 |
sebras | chrisl: it has an error callback that differentiates between the levels, is that nice enough? :) | 16:51.17 |
chrisl | Ah, did not know that.... | 16:52.03 |
sebras | chrisl: base/sjbig2.c implements that callback but only prints something if jbig2dec emits fatal. | 16:52.45 |
| warnings seem to be passed to if_debug3m('w', ...') | 16:53.00 |
| but I'm not quite sure how to get those printed, even in a debug build. | 16:53.10 |
chrisl | -Zw in a debug build | 16:53.26 |
| sebras: If you want to rejig s_jbig2decode_error() to also print JBIG2_SEVERITY_WARNING messages on a non-debug build, I certainly wouldn't complain | 16:55.25 |
sebras | chrisl: jbig2dec now prints a fatal error and a series of warnings so it is easy to spot where the error originated. | 16:56.57 |
| chrisl: and if there is a problem it can cope with that is also a warning (but no error) | 16:57.15 |
| chrisl: seems sane..? | 16:57.22 |
| just for context, this is an example of a file where jbig2dec recovers: | 16:59.11 |
| https://pastebin.com/raw/ZmdNArFm | 16:59.12 |
chrisl | That looks fine to me | 17:02.03 |
| It is generally our preference to tell people there's a problem, even if it is a problem we can continue from | 17:02.43 |
| Okay, heading off...... | 17:07.49 |
sebras | chrisl: np, thanks for the help. | 17:08.04 |
| Forward 1 day (to 2018/07/24)>>> | |