Re: [emacs-pms] Introduction to the Emacsmirror

[Tom Breton (Tehom)]

[Another message I mis-addressed the first time.  I've updated ]
>> * Emtest at http://savannah.nongnu.org/git/?group=emtest
>
> In order to extract metadata a package I need to know the "main" file of
the package. In most cases a package foo has a main file "foo.el". This
is automatically detected. In some cases I have to manually specify the
main file, e.g. for your org2blog I would use "main.el". But this
package does not seem to have a main file at all.

Right, there really isn't a main file.  There's a main loop, but that's
not an entry point.  The entry points are plugins.  I recently reorganized
the directory structure.

I suppose I could make a pseudo-main file to satisfy your requirements. 
What do you need in a main file?

Or I could just write the metadata for you, if you prefer.

> I have imported ~2900 packages and it is a problem for me when people do
unusual things. This doesn't mean I am against trying new things - I am
just saying that it can cause problems for me (and possibly others) and
that you should keep in mind that following the
> conventions also has benefits.
>
> Another such problem with your libraries are your ;;;_ prefixes, and
similar. My script won't find the library summary and commentary because
of this.

They are allout outline headings.  I organize my files outline-style, and
allout is the only outliner I found that co-operates with source code.  At
least it certainly was when I started and when I have rechecked it still
seems to be.

Some use folding mode.  Would that solve the problem for you?  Is your
script able to process folded code such as g-client, eukleides.el or
moikrug.el?  Or folding.el itself?

Since sending this the first time, I've had another look at folding mode. 
I can see maybe switching to it if it solves your problem, but I can't
promise.

> Here again you are not following the conventions and
> contrary to the above issue where I think it is worth experimenting I
think that you should not do it. For section separators I do not care as
there already exist many different ways in which people encode nesting
and other things in these title lines, so that I can not parse them
anyway. But for the header you should follow the respective conventions.
>
> You probably have some library that makes use of the information by your
various library prefixes but it just makes it impossible for any other
tool to parse your comments. I won't adapt my import scripts to support
your variation, all the hacks required to accommodate the various
non-standard headers in various packages would just sum up and add much
complexity to the import scripts. Given the ~2900 packages I mirror my
only option in such cases is to simply ignore the issue and have
incomplete metadata on the mirror.



>> I also have git://repo.or.cz/org2blog.git. Â I didn't realize I hadn't put
>> it on a repo until I was answering you. Â It's up now.
>
> I am already mirroring this org2blog:
> https://github.com/punchagan/org2blog
> Maybe one of you can change the name.

Since then, I have talked to Puneesh.  We've agreed that mine will be
org2blog/atom and his org2blog/wp.

Do you have any difficulty mirroring it as "org2blog/atom" or
"org2blog-atom"?
It's still at git://repo.or.cz/org2blog.git

Another package of mine:
 * tinydb at git://repo.or.cz/tinydb.git

Best of luck,

        Tom Breton (Tehom)