1. trace file
- Posted by useless_ Sep 23, 2013
- 1552 views
Can some links be added to the help files concerning trace, ex.err, and crashing? All i wanted was to redirect where the ex.err file was written, and it took an hour to re-discover how to do it. The docs on crash_file() seem to be saying "use this to crash the program", trace docs don't mention redirecting the trace file when it crashes, there's no search results for "ex.err", there are no returns for "specify crash file", no returns on "crashfile", etc etc etc.
useless
2. Re: trace file
- Posted by DerekParnell (admin) Sep 24, 2013
- 1523 views
Can some links be added to the help files concerning trace, ex.err, and crashing?
A good idea. I'm sure we can add better docs for this important area.
The docs on crash_file() seem to be saying "use this to crash the program" ...
I'm not seeing that. I think the docs on crash_file() are quite clearly saying that all it does it nominate where crash messages go, in the event that the program happens to crash for whatever reason. It never says anything like "use crash_file() to cause your program to crash. Maybe you have confused the crash() routine with this instead.
3. Re: trace file
- Posted by useless_ Sep 24, 2013
- 1466 views
Can some links be added to the help files concerning trace, ex.err, and crashing?
A good idea. I'm sure we can add better docs for this important area.
The docs on crash_file() seem to be saying "use this to crash the program" ...
I'm not seeing that. I think the docs on crash_file() are quite clearly saying that all it does it nominate where crash messages go, in the event that the program happens to crash for whatever reason. It never says anything like "use crash_file() to cause your program to crash. Maybe you have confused the crash() routine with this instead.
But it is a procedure, procedures do something. To me, calling crash_file() says to go write a crash_file right now. But i don't want a crash_file right now, even if i want a file containing that same data for debugging at this point, i don't want a crash.
I kept trying to set (variable) crash_file = location.
useless
4. Re: trace file
- Posted by ghaberek (admin) Sep 25, 2013
- 1412 views
But it is a procedure, procedures do something. To me, calling crash_file() says to go write a crash_file right now. But i don't want a crash_file right now, even if i want a file containing that same data for debugging at this point, i don't want a crash.
I kept trying to set (variable) crash_file = location.
useless
crash_message() and crash_routine() are the same way. It does seem a bit ambiguous. These should probably be called set_crash_file(), set_crash_message(), and set_crash_routine() so that they include verbs. And if we're picking everything apart, warning_file() should be set_warning_file(), and warning() should be warn() or issue_warning().
-Greg
5. Re: trace file
- Posted by _tom (admin) Sep 26, 2013
- 1378 views
In std/error.e why is the language usage inconsistent?
An error condition is described by:
- crash_message
- crash_file
- crash_routine
- crash
- ex.err
Language switches from "error" to "crash" back to ".err".
While a "warning" condition is uniformly described:
- warning_message
- warning_file
- warning
The process for an "error" would be better described by: error_message, error_file, error_routine, error, ex.err
We could use exception to mean either error or warning in the same way routine is used to describe procedure or function.
The result would be:
- make it simple, make the same
- everything fits into a uniform pattern
- easier to document
- easier to remember and use
_tom
