Anton Lindqvist wrote:

> > > If a file passed to the 'qf_init_ext' function contains lines longer
> > > than 1021 bytes (might be platform specific) the rest of the line is
> > > recognized as a separate error without a valid filename and lnum. If a
> > > string or list is passed to 'qf_init_ext' the rest of a long line is
> > > discarded. This patch adds the same behavior while reading errors from a
> > > file. This is done by continue reading from the file in to a temporary
> > > buffer until end-of-line is encountered. Could any existing general
> > > purpose buffer with a known size be used instead of allocation a new
> > > one?
> > 
> > This code comes from when memory was scarce and error messages were
> > short.  Using a fixed size buffer isn't so nice these days.
> > 
> > A good strategy could be to use IObuff as it is, and when an error
> > message is longer then allocate a larger buffer and append to it.
> > It does require growing (doubling?) the buffer further until a NL is
> > found.  That can be slow, e.g. if the error file has no NL at all
> > (that's a mistake, but you don't want Vim to hang then).  So some upper
> > limit would still be useful.
> 
> I would prefer allocating a buffer using BUFSIZ defined in stdio.h and
> then make it grow exponentially. Instead of using a upper limit I would
> prefer to continue reading until NL/EOF occurs or realloc fails.

Well, if arbitrary long lines are supported, where would they be useful?
Certainly not for displaying, it would scroll through many pages.
Perhaps for a script that analizes the error further.  But I find that a
bit far fetched.  I think a 100 Kbyte limit is very resonable.
Especially since currently it's chopped up after 1 Kbyte.

> > I have seen error messages from a test that say "expected
> > <very-long-string> but got <other-long-string>".  Not sure what a useful
> > upper limit would be.  100 Kbyte perhaps?
> 
> Reading and displaying the whole line despite its length is a reasonable
> behavior nowadays, however like marius@ pointed out I would not mind if
> the lines where truncated since they seldom adds value. 
> 
> > Are you interested in implementing this?
> 
> Sure! But what's your take on the concerns above?

I think reading into IObuff at first, and using a growing buffer after
that would be a good solution.  Probably keep the buffer until all
errors are done, if one line is long there is a bigger change of another
one.

-- 
hundred-and-one symptoms of being an internet addict:
229. You spend so much time thinking what to add on this list.

 /// Bram Moolenaar -- [email protected] -- http://www.Moolenaar.net   \\\
///        sponsor Vim, vote for features -- http://www.Vim.org/sponsor/ \\\
\\\  an exciting new programming language -- http://www.Zimbu.org        ///
 \\\            help me help AIDS victims -- http://ICCF-Holland.org    ///

-- 
-- 
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php

--- 
You received this message because you are subscribed to the Google Groups 
"vim_dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Raspunde prin e-mail lui