Trace #1332
Trace #1332
Conversation
This adds functionality for tracing internal functions by manually adding two function calls to each function. The first call takes an argument which is intended to identify the function being called and (possibly) the arguments with which it has been called (in testing some arguments don’t take well to outputting to the log file, and so are omitted). The second call takes no argument and serves to indicate the end of the function. While the output of these functions is interwoven into the normal log file, it has a unique prefix (GreTrace) allowing filter tools to quickly isolate these lines. We also make use of indentation in order to indicate the nesting of function calls. Output is formatted as follows: ``` GreTrace: gre@generatelines GreTrace: ----- GreTrace: gre@removespaceabove GreTrace: | gre@generatelines GreTrace: | ----- GreTrace: ----- GreTrace: gre@generatelines GreTrace: ----- GreTrace: gre@updateleftbox GreTrace: | gre@updatelinewidth GreTrace: | | gre@generatelines GreTrace: | | ----- GreTrace: | ----- GreTrace: ----- GreTrace: gre@widthof GreTrace: ----- ```
I noticed that there was code in gregoriotex-main.tex which had to be duplicated in gregoriosyms.sty in order for that functionality to work properly in the symbols only style. However, some recent changes in gregoriotex-main.tex weren’t reflected in the code in greogiosyms.sty resulting in an inconsistency. Rather than simply fix that inconsistency, I’ve put the code that is supposed to be common to both in a separate file and included that file in both locations. This way there’s only one place where the changes have to be made and they’ll be reflected in both uses.
Some arguments do not take well to expansion attempts before being written to the log and by not expanding the arguments what appears in the trace log is the function as it was called in the code, making it easier to find the right code location for an error (which is the whole point of this functionality). Therefore, we prevent `\gre@trace` from expanding its arguments.
I’m not creating a CHANGELOG entry for this as this is entirely internal.
|
Why aren't all macros (i.e., user and gregorio macros) traced as well? |
|
I only traced the internal macros because the calling of them tends to be more hidden than the calling of user or Gregorio namespace ones, which actually appear in source files (or gtex ones). I could add tracing to all macros if that would be more useful. |
|
I guess some finesse is needed. It's probably important to trace user and Gregorio macros that are called from internal macros. |
|
That is probably a good point, but I don't know how to design a mechanism to selectively track such macros based on who is calling them. |
|
Well, I just meant being selective in which user and gregorio macros are traced. |
|
That would be substantially easier to pull off. Do you have a list of macros which you think should be traced? |
|
Based on recent experience, |
* develop: Documentation Set up the left box using next line heights in \GreDiscretionary. Fixes gregorio-project#1285. Documentation correction Fixed infinite recursion. Implemented function to set adjust line height for next line. Part of the implementation for gregorio-project#1285. Oops. Forgot to stage a change. Swap some \hskip to \gre@hskip More redundant function calls More redundant function calls Move file version declaration Remove useless debug messages Removing redundant function calls
|
Okay, I've added tracing to all of the possible functions in gregoriotex-signs.tex. Take a look at what that does, and let me know if I should do the same for the other files. |
|
I think this is good enough. More can be added as we encounter the need to add it. |
I started this while working on 1285 and found it so useful that I went whole hog and formalized it and applied it to all the internal functions. It could be applied to Gregorio or user name space functions as well, but I'm less sure that there's a gain from that.
Nominally at least, tests aren't supposed to change as a result of this PR, but #1331 was found as a result of applying this, so that needs to be resolved before this is merged.