Re: VistA packaging : vista-foia : First review

Hi Andreas,

On Sun, Feb 2, 2014 at 10:33 AM, Andreas Tille <andreas@an3as.eu> wrote:

I have done some `git checkout pristine-tar` in several cases in
repositories and found pretty small files since it only cares for
metadata not for actual content files. So these files should definitely
not be large.

Yes, I must be doing something wrong in the way I'm attempting to

import the tar file.

Maybe the problem is related to the package name.

The .orig.tar.gz file created by git-buildpackage contains the base

directory:

vista-foia-0.20130405

with the version included.

Is that the standard practice ?

Should that matter ?

>
> Yes, it does have some large files.
>
> Mostly the *.zwr files, that contain the MUMPS Globals. they
> get very large. For example, 26 of them are 64Mb each,
>
> https://github.com/OSEHRA/VistA-M/tree/master/Packages/Lexicon%20Utility/Globals
>
>
> In total there are 2,933 of those .zwr files.
> The majority are rather small, but... it adds up in the final tar file.

As mentioned above this should not be any problem regarding creating
pristine-tar. However, once we plan to upload this to the Debian mirror
we will not be able to circumvent the question of our ftpmaster: "Where
is the source of this binary chunk of data and how can it be
reproduced."

We should be able to provide a reasonable answer to this ... even better
if we could somehow recreate the data in the build process (if this is
any possible and makes any sense at all).

My bad,

I missed to clarify that these are not binary files.

They actually contain plain text.

For example:

https://raw2.github.com/OSEHRA/VistA-M/master/Packages/Lexicon%20Utility/Globals/757+MAJOR%20CONCEPT%20MAP.zwr

Has a large number of entries such as :

...

^LEX(757,62,0)="332^"
^LEX(757,63,0)="334^"
^LEX(757,64,0)="340^"
^LEX(757,66,0)="351^"

...

So, they are still "source" files.

> Regarding the difference in behavior of git-buildpackage,
> one potential reason might be the values in the configuration files
> http://honk.sigxcpu.org/projects/git-buildpackage/manual-html/man.gbp.conf.html
>

As I said in my other mail I'm using the configuration file suggested by
Debian Med policy (and used by several other teams as well for the
reasons I mentioned in my other mail).

Yes, that sounds reasonable,

and now that it is clear that the content of the branch

should a rather small file, there is no reason why we

couldn't get it right.

I just need to figure out what I'm doing wrong with the

directory names, or directory locations... I'm most

likely missing a very basic detail....

I'll keep looking.

Thanks

Luis