Re: Regtest compiling

[James Bailey]

On Nov 30, 2010, at 7:47 AM, Graham Percival wrote:

> On Tue, Nov 30, 2010 at 07:34:04AM +0100, James Bailey wrote:
> 
>> 3. How do tracking memory usage and source code changes get me any closer to 
>> easily compiling the regtests? I'm guessing they probably don't.
> 
> Sorry, I don't understand this?  Typing "make test" will compile
> the regtests; I thought that was what you wanted?

Ah, that's really useful information. I didn't know that.
> 
>> Maybe there needs to be a separate section in the CG on that occasional task 
>> of manually compiling all of the regtests?
> 
> ...
> I don't think anybody should be manually compiling regtests, but
> you could do something like:
> 
> cd input/regression/
> for f in *.ly; do lilypond $f; done;
> 
> Note that in this case, you don't even need to compile lilypond
> itself.  All you need is the git repository, and a normal GUB
> lilypond.

True, but like I said, I'm trying to come as close as possible to mimicking the 
actual user experience of clicking on the image in the html documentation, 
copying everything between the copy/paste sections, and running lilypond on it. 
That helps me a lot when trying to understand a particular aspect of something, 
whether it's in the regtests or in the documentation. And two things compiling 
every example in the regtests taught me were that there are really useful 
concepts in there which I don't currently use but could, and that when checking 
that the output of a particular regtest is supposed to be on a set number of 
pages, or that the margins should be in a particular place, there's no real 
substitute for actually compiling the regtest.
> 
>> Basically, I want a way to have access to the regression tests
>> so without the header declaration at the top so that I can
>> easily run lilypond on them. I'm sure, since I can compile the
>> documentation, whatever is used to generate the regression tests
>> is there somewhere, but since I am a noob, I do things the only
>> way I know how. which is to copy all of the *.ly files into a
>> folder, remove the first couple of lines from all of them, and
>> run lilypond on them.
> 
> Why remove any lines?  The compiled version of the regtests have
> something added, but the version in git can be easily compiled on
> the command-line.
> 
exactly, if I just compile the exact regtest with the \include 
"lilypond-book-preamble.ly", it errors with the following:
 Processing `/Users/jamesbailey/Desktop/lilypond snippets/lily-fee7c63f.ly'
Parsing...
Renaming input to: 
`/main/src/gub/target/linux-x86/src/lilypond-git.sv.gnu.org--lilypond.git-release-unstable/input/regression/accidental-broken-tie-spacing.ly'
Interpreting music... 
Preprocessing graphical objects...
Calculating line breaks... 
Drawing systems... 
error: 
The PostScript backend does not support the
system-by-system output.  For that, use the EPS backend instead,

  lilypond -dbackend=eps FILE

If have cut & pasted a lilypond fragment from a webpage, be sure
to only remove anything before

  %% ****************************************************************
  %% Start cut-&-pastable-section
  %% ****************************************************************

commenting out, or removing that line, which is something I also do with sed, 
solves that problem

Also, what is this version in git you speak of? I'm sure I have it, I just 
don't know what I'm looking for.

> I'm not familiar with the specific problem you're trying to find,
> so I can't speak to that.

As I said, I discovered last year a problem that only ocurred when the regtest 
was actually compiled to a pdf and looked at, namely that the margins were off. 
I can't predict what other problems could occur, but something like that is 
easy to see in a pdf where nothing has been changed, and difficult to see in 
html where the margins are different.