Debugging byte compiled files

Debugging byte compiled files

[Phillip Lord]

I've recently found a bug in my own pabbrev.el package. It's taken
me so long because it only occurs when the file is byte compiled
and I don't routinely byte compile my own packages. 

I think that the problem is coming from this macro...

(defmacro pabbrev-save-buffer-modified-p (&rest body)
  "Eval BODY without affected buffer modification status"
  `(let ((buffer-modified (buffer-modified-p)))
     ,@body
     (set-buffer-modified-p buffer-modified)))

the backtrace looks something like this...


(macro . #[(&rest body) "\301\302\303\b\304\"BB\207" [body let
((buffer-modified (buffer-modified-p))) append ((set-buffer-modified-p
buffer-modified))] 5 ("/home/oub/emacs/site-lisp/versch/pabbrev.elc" .
17814)])

Backtrace is: 
  backtrace()
  pabbrev-command-hook-fail((invalid-function (macro . #[...
"\301\302\303\b\304\"BB\207" [body let ... append ...] 5 ...])) "pre")
  pabbrev-pre-command-hook()
  run-hooks(pre-command-hook)


I have no real clear idea how to debug this. The error happens
frequently
but not repeatedly, and it all happens in the pre-command-hook. It has 
never occurred to me without byte compilation. 

Has anyone got any idea how I go about finding out what the problem is? 

Cheers

Phil

Re: Debugging byte compiled files

[John Paul Wallington]

"Phillip Lord" <Phillip.Lord@newcastle.ac.uk> writes:

> I've recently found a bug in my own pabbrev.el package. It's taken
> me so long because it only occurs when the file is byte compiled
> and I don't routinely byte compile my own packages. 
>
> I think that the problem is coming from this macro...
>
> (defmacro pabbrev-save-buffer-modified-p (&rest body)
[...]

Does the `defmacro' form occur before its use within your library ?

Re: Debugging byte compiled files

[Stefan Monnier]

>   pabbrev-command-hook-fail((invalid-function (macro . #[...

This class of errors means invariably that a call to macro F was
byte-compiled while Emacs didn't know that F was going to be defined as
a macro rather than a function.
I.e. the macro didn't exist yet.
Either because it's defined later in the file, or because of a missing
`require'.


        Stefan

RE: Debugging byte compiled files

[Phillip Lord]

John Paul Wallington wrote:
> "Phillip Lord" <Phillip.Lord@newcastle.ac.uk> writes:
> 
>> I've recently found a bug in my own pabbrev.el package. It's taken me
>> so long because it only occurs when the file is byte compiled and I
>> don't routinely byte compile my own packages.
>> 
>> I think that the problem is coming from this macro...
>> 
>> (defmacro pabbrev-save-buffer-modified-p (&rest body) [...]
> 
> Does the `defmacro' form occur before its use within your library ?


There is a single use of the macro before its use. You question
suggests to me that this is a bad thing?

Phil

Re: Debugging byte compiled files

[John Paul Wallington]

> >> I think that the problem is coming from this macro...
> >> 
> >> (defmacro pabbrev-save-buffer-modified-p (&rest body) [...]
> > 
> > Does the `defmacro' form occur before its use within your library ?
> 
> 
> There is a single use of the macro before its use. You question
> suggests to me that this is a bad thing?

Yup.  Sorry for not being clear.  Stefan has already explained the
problem really well.  See also (elisp) Compiling Macros:

"In order for compilation of macro calls to work, the macros must
already be defined in Lisp when the calls to them are compiled.  The
compiler has a special feature to help you do this: if a file being
compiled contains a `defmacro' form, the macro is defined temporarily
for the rest of the compilation of that file.  To make this feature
work, you must put the `defmacro' in the same file where it is used,
and before its first use."

RE: Debugging byte compiled files

[Phillip Lord]

Stefan Monnier wrote:
>>   pabbrev-command-hook-fail((invalid-function (macro . #[...
> 
> This class of errors means invariably that a call to macro F was
> byte-compiled while Emacs didn't know that F was going to be defined
> as a macro rather than a function. I.e. the macro didn't exist yet. 
> Either because it's defined later in the file, or because of a
> missing `require'. 
> 


There you go. Learn something every day. I didn't think that the
order made any difference. 

I'd put you in the acknowledgements of pabbrev for helping me out, 
but you're already there!

Cheers

Phil