Completion of error handling

[elfring]

Would you like to add more error handling for return values from functions like the following?

• fclose ⇒ midi_write
• fprintf ⇒ write_font_stream

[PhilipHazel]

The PMW code base is over 30 years old. In all that time I have never had any error problems with fclose() or fprintf(). There are a LOT of calls to fprintf(). Checking the return code each time would use quite a lot of computer cycles. Besides, PMW is just a user program with no privileges; if it does crash it can't do any harm.

[elfring]

How do you think about to improve static source code analysis also for this software? 🤔

[PhilipHazel]

I will look at any issues that people send me, but this is old code with a small number of users, and I currently have no plans for any major work on it.

[elfring]

👀 There are development tools available which can influence software evolution in more desirable directions.

[PhilipHazel]

Indeed, but I am far too old to learn these new tricks.

[elfring]

💭 I suggest to avoid ignorance of return values a bit more.
Would you like to detect every error situation as early as possible?

🔮 Under which circumstances will you care more also for data output failures?

[spaskalev]

Indeed, but I am far too old to learn these new tricks.

Regardless of its age PMW is quite a valuable piece of software. I understand where the push to modernize it comes from though. Better distributions that may want to package it, as a rule, run security scanners on the packages that they build and distribute. The more issues found and fixes, the better in the end. PMW may be user software but I've planned for a while to make it run in the browser via cross-compilation or as a sort of cgi endpoint - again so that users that only have a web browser can use it without installing it on their devices (e.g. think tablet users or non-technical users like musicians) If this is done pmw is no longer the end-user software that doesn't affect anything if it crashes.

[PhilipHazel]

My goodness, PMW is more widely used than I thought. I take your point; I have in any case been thinking this issue over and may introduce some more checks.

[elfring]

My goodness, PMW is more widely used than I thought.

🔮 The application popularity might grow more.

… I have in any case been thinking this issue over and may introduce some more checks.

Are there chances that the awareness will also grow for cross-cutting concerns? 🤔

[PhilipHazel]

I have added error checks to fclose() and to fprintf() except when it is writing to stderr. The important cases are when the return value of fprintf() is actually used (there are not many).

[elfring]

Thanks for your source code adjustment.
🔮 How will chances evolve to benefit any more also from the means of aspect-oriented software development?

---

Another update candidate:

pmw/src/midi.c

Line 1076 in 201987c

fseek(midi_file, 18, SEEK_SET);

[elfring]

How do you think about to take also another look at return values from fwrite() calls? 🤔

[elfring]

👀 Further update candidates:

pmw/src/pdf.c

Lines 3069 to 3083 in 201987c

	(void)fprintf(out_file, "xref\n0 %d\n", objectcount);
	(void)fprintf(out_file, "0000000000 65535 f\r\n");
	for (pdfobject *p = obj_anchor; p != NULL; p = p->next)
	fprintf(out_file, "%010lu 00000 n\r\n", p->file_offset);
	/* Now write the trailer, which contains (inter alia) the number of objects and
	where to start. */
	fprintf(out_file, "trailer\n<</Size %d/Root 1 0 R/Info 2 0 R\n", objectcount);
	fprintf(out_file, "/ID[<%s><%s>]>>\n", md5ID, md5ID);
	/* Finally, the offset to the start of the xref table. */
	fprintf(out_file, "startxref\n%u\n%%%%EOF\n", filecount);

[PhilipHazel]

Thank you for noticing those cases that I had overlooked.

[elfring]

👀 Thanks for corresponding source code adjustments.