[Debian-science-sagemath] GAP: issue related to compressed manual.six: PATCHES

[Jerome BENOIT]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Hello Forum,

On 06/12/16 15:34, Ximin Luo wrote:
> Jerome BENOIT:
>> Hello Bill, thanks for your comment.
>> 
>> On 06/12/16 10:44, Bill Allombert wrote:
>>> This is too intrusive and does not solve any issue with gap.
>> 
>> It is intrusive, I am agree. But it also solve the issue. I do not
>> think that a mature software may use piping.
>> 
> 
> Why do you say this? Piping may be a bit untidy on the systems level,


> but on an abstract level it's a good way to *compose* arbitrary
> programs together, that are written in different programming
> languages, in a *generic* way.

using available libraries is also a generic way: the patch is generic in the
sense that it just plugs the zlib library (other compressor libraries can
be plugged in). GAP is mature, not in development phase.

> 
> I think the principle of trying to write everything in one language,

zlib exists and is used by a large family of software, so the `principle of trying'
is not applicable here. From a practical point of view, especially when you are
dealing with insane linux boxes (quite common in labs), have your software written
in C saves your day.

The solution are not equivalent since zlib allows to management compressed files
as (custom) streams. In other words seek can be used, while piping does not allow it.
Here, for GAP, this avoid to patch (badly) the material written in GAP language.
Furthermore, exposure to the EPIPE mess is avoided: dirty scripts involing GAP
can be written without fear.


> is slightly misguided. I think it's reasonable for GAP to use gzip
> instead of zlib. For example if they wanted to switch to xz
> compression, they can just replace the program name.

and breaks backward compatibility along the way.


 With an API,
> they would need to rewrite all the API calls.

if you replace zlib by libzio,  you just delegate the management of
files compressed with different compressor to libzio. 

> 
> Anyway, we don't need to argue about software engineering philosophy
> here. Concretely, we have potentially a 1-line patch to Sage to fix
> the issue. And upstream python have already fixed it in 3.5. I will
> see if I can test it tonight.

Let do this change. Note that it must be also applied to libgap tests.
Second, the gap and libgap packages at deb-sci-sage are patched and
the doctests related to GAP and libGAP do not failed with them.
In short, you must do the tests with GAP and libGAP as currently furnished
in unstable. (I may apply the patch to the libgap-sage package, and I may
consider to submit them upstream at least to show that they can easily
deal with compressed material).

Thanks,
Jerome


> 
> X
> 

- -- 
Jerome BENOIT, Ph.D. | jgmbenoit-at+rezozer*dot_net
http://www.rezozer.net/

- -- 
Jerome BENOIT | calculus+at-rezozer^dot*net
https://qa.debian.org/developer.php?login=calculus@rezozer.net
AE28 AE15 710D FF1D 87E5  A762 3F92 19A6 7F36 C68B
-----BEGIN PGP SIGNATURE-----

iQQcBAEBCgAGBQJYRvjEAAoJED+SGaZ/NsaLe00f/07imdJIxHtD3zC7VceVoGVZ
D7AHHUWCsAIKNocMyq57l8nCk6HSF22CgnlZaQJtefzwADB+WUMkPSXAh9iVcnEV
hpe/wrW62LnUe17xQn6be8wcc+oibOFY4K2KWFuc7WyTcvQOZ/znpGQwZa6ft4fA
NejiErvHwq+2vwTBJFajZfXGCain1RnQp24CL0gv4p26zsVaIT20F90p/EBUKAKm
lWr+Eaggtth9ZfjonmnOis0/fxJ48iAuqAoo0tH9R+EBKCiu17lxfr2QjdEITenX
XoUPSi1YrjFaGP88EcC7eriWu5krtUTfBFu2+R4gXn6HP0E+8T7zNs3/aoDenQ8o
APMdCts4hMH1kjLSjoQNgNJyxhNZagRtLyuHZQg8JTLI41E7NPmWdCF49pGz0Nbh
Kpcx/AAj9K1+/JUz3GpvEY1J05zMZvvkilEs7fyRDU1hL1w1jY2RSJL/75ic3AcM
FRkdMYSIENl+WBYCume4xIhVUDDEpVoYsuAqo4bgEo2M+gvurdYrGw+fSUHCqo0g
S1ZMI+AYUFBJtZhrqVofcXYrvAhHTmUXSeOmPNI+NXHkunA0hiRMfrQ5ifNodewv
g+vPWICMIIbRl9o8Nb9WrcFqnBdJlu2GG8zSdi1varyC9k0oIrg9dRZ99ZBY4lBu
ZvEpv1T8/2yBvX5RnlwaE/0QLfDwC6PIE64kTHhW9YclcZnr3557iKDmZkoNUhW6
TVNAFjLcSvuGNpEO0u6+JOvhqsSQrKXgLqbkBWLm4qOngPhBrYCKLkQK8EcUmJ71
7JNGE+ZrMfK5tuDDB7SPgAA0oP6qucus3tU85Z7VsqGyoawITtx15jXa1GjjGM1Q
JPNLzVFyqmYp9gw50iWE+w0u5oIjUoNcXqAUm4DDEOa07erb5KBYVyKDRfDTzqmW
pME6uoPQWLD9Q3lt0H0kPv0ryYmd5Otw5su9NESlTE/WCuzVvp1N/3O9NhIkn9kh
71YJKUj76o3aAYSk8c56kwIDGdWfeghrPSvcP22aewLZ+fQrVWwCSXKdbsTQm5zu
Q+m/fIQY4uetz6vZ6CJkQq5ht+Dj2llJESozvj000gCN4YeI1VeGv++RIdpBhOIp
mIrgTT+pRVkzkjccYAspX6eViy9wJ7scrp0HLeN7kSQl6LJxkYCS1uu3goFgM4W5
3zDzNtJyZmZxyenPTQ6YB6rDVXRgvDPmei4cmoroq9A0B/9+gkAbOUD5cweIIZXL
/yOO0CvlgWM3T2I2AAS0ORXwwN92i/YwyvmxnL4Ma8/vnrpLn5GESXn8+3/UfWym
2ECznx2+MHZPskiruaL8gmzpZ5ub1pdMdPGafMdkuQTSi6iXSnQ6kCbk4Diop7Q=
=OrSw
-----END PGP SIGNATURE-----