emacs-orgmode
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Orgmode] Re: [CEDET-devel] Org-mode


From: Eric Schulte
Subject: Re: [Orgmode] Re: [CEDET-devel] Org-mode
Date: Thu, 04 Nov 2010 13:16:03 -0600
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/24.0.50 (gnu/linux)

Jambunathan K <address@hidden> writes:

> The main challenge with breaking up org-html.el first and then say
> plugging in org-odt.el to that later is:
>
> 1. Code-churn that it would create in org-html.el
> 2. Proving that *nothing* in HTML export actually breaks.
>
> Carsten would like to avoid (1) - he might want to go with one bit
> commit and (naturally) shift the responsibility of (2) to the committer.
>
> So a committer has one more thing he needs to be concerned about.
>
> I do see some regression tests for html exporter and I am unsure how
> *complete* they are. What would have really made things easier is the
> following:
>

I don't believe that there are any existing unit tests for the org html
exporter, however writing a suite of such tests seems like the best
first step in this project.  It would allow sanity checks throughout the
implementation of the parser, and would give some confidence in the
parsers functionality before merging back into the core.

Org-mode has a test suite based around ERT, which should facilitate this
process.  Look in the testing directory of the Org-mode repository.

>
> 1. One Org file in repo that has *all* the Org-specific markups.

Yes, either one large file or many smaller files.  For the Babel tests
we use one large file with one subtree per test s.t. it is easy for a
specific test to jump to the appropriate section of the file.

> 
> 2. One HTML file that is a exported from this Org file which is
> re-checked in as and when the org-html.el changes it's markups.
>
> Everytime something changes in the exporter one just diffs the *new*
> HTML file with the one in the repo and be able to say with assured
> confidence that something has improved or broken.
>

I think a series of feature specific unit tests would provide more
granular feedback, would allow incremental development, and would allow
differences in non-essential properties like white-space.

Best -- Eric



reply via email to

[Prev in Thread] Current Thread [Next in Thread]