sagemath/sage

Modify giac upstream tarball to remove non-portable filenames

mkoeppe opened this issue · 37 comments

http://www-fourier.ujf-grenoble.fr/~parisse/debian/dists/stable/main/source/

When preparing the reduced tarball with spkg-src,
we should rename the non-portable file names with a colon in them:

$ find . -name "*:*"
./doc/pari/Programming_in_GP:_control_statements.html
./doc/pari/Programming_in_GP:_other_specific_functions.html
./doc/pari/cont_Programming_in_GP:_other_specific_functions.html
./doc/pari/cont_Programming_in_GP:_control_statements.html

(problematic at least on Windows - as reported in #28549.)

Previous updates:

Follow-up: #29552 (Upgrade giac to 1.5.0-87) ... for Sage 9.2

Upstream: Fixed upstream, in a later stable release.

CC: @antonio-rojas @dimpase @frederichan-IMJPRG @sagetrac-parisse @slel @infinity0 @saraedum @tobihan @timokau @kiwifb

Component: packages: standard

Author: Matthias Koeppe

Branch/Commit: a9b4666

Reviewer: François Bissey, Dima Pasechnik

Issue created by migration from https://trac.sagemath.org/ticket/29521

Description changed:

--- 
+++ 
@@ -1 +1,16 @@
 http://www-fourier.ujf-grenoble.fr/~parisse/debian/dists/stable/main/source/
+
+When preparing the reduced tarball with `spkg-src`, 
+we should rename the non-portable file names with a colon in them:
+
+```
+$ find . -name "*:*"
+./doc/pari/Programming_in_GP:_control_statements.html
+./doc/pari/Programming_in_GP:_other_specific_functions.html
+./doc/pari/cont_Programming_in_GP:_other_specific_functions.html
+./doc/pari/cont_Programming_in_GP:_control_statements.html
+```
+
+(problematic at least on Windows - as reported in #28549.)
+
+

Description changed:

--- 
+++ 
@@ -13,4 +13,7 @@
 
 (problematic at least on Windows - as reported in #28549.)
 
+Previous updates:
+- #26315 Upgrade to Giac 1.5
+- #28101 update giac to 1.5.0-63
 

Upstream: Not yet reported upstream; Will do shortly.

comment:5

Would be nice to do this. There are a few trivial doctests impacted by this. But I'll wait to know what upstream intend to do about the naming before going ahead.

Changed upstream from Not yet reported upstream; Will do shortly. to Reported upstream. No feedback yet.

comment:7

This is a problem that should be reported to PARI:
https://pari.math.u-bordeaux.fr/dochtml/html-stable.tgz

comment:8

Replying to @sagetrac-parisse:

This is a problem that should be reported to PARI:
https://pari.math.u-bordeaux.fr/dochtml/html-stable.tgz

Thanks for that . Which now begs the question, should we install pari's html doc with giac? We already have the documentation which comes with pari, in pdf if I am not mistaken.

comment:9

The HTML doc is intended to be used from Xcas (Help, Manuals, PARI-GP), I don't think you will need it to run giac inside sage.

comment:10

xcas is the fltk application, right? I am not sure we ever build it in sage as we would have to make sure there is a proper fltk install and that it is detected properly. So, not shipping those won't hurt I guess.

comment:11

Yes xcas is never built by the spkg because we put in spkg-install:
sdh_configure --disable-gui
I did this long ago to prevent the configure script to find a bad version of fltk that might be installed on the system.

comment:12

OK sounds like our just spkg-src needs to delete more stuff.

comment:13

Replying to @sagetrac-parisse:

This is a problem that should be reported to PARI:
https://pari.math.u-bordeaux.fr/dochtml/html-stable.tgz

apparently it's already fixed, for the upcoming release, i.e.
https://pari.math.u-bordeaux.fr/dochtml/html.tgz
I see no : in filenames.

For html-stable, I see


-rw-r--r-- 1 dimpase dimpase   26979 Jul 19  2016  Programming_in_GP:_control_statements.html
-rw-r--r-- 1 dimpase dimpase   39993 Dec  1  2018  Programming_in_GP__control_statements.html
-rw-r--r-- 1 dimpase dimpase   46872 Jul 19  2016  Programming_in_GP:_other_specific_functions.html
-rw-r--r-- 1 dimpase dimpase   78935 Dec  1  2018  Programming_in_GP__other_specific_functions.html

so these files with : are just leftovers, by looking at dates...

comment:14

A decision needs to be made whether for 9.1, we want to do the update to to 1.5.0-87 or just clean up the tarball of 1.5.0-63

comment:16

Would someone who knows about the spkg-src be willing to update it?

comment:17

Replying to @mkoeppe:

Would someone who knows about the spkg-src be willing to update it?

I'll admit that I don't really want to touch it. I would probably have wanted to do the whole thing slightly differently.

Commit: 31d2b93

Description changed:

--- 
+++ 
@@ -17,3 +17,4 @@
 - #26315 Upgrade to Giac 1.5
 - #28101 update giac to 1.5.0-63
 
+Follow-up: #29552 (Upgrade giac to 1.5.0-87) ... for Sage 9.2

New commits:

036dbbabuild/pkgs/giac/spkg-src: Remove files with nonportable names and other unneeded files
31d2b93build/pkgs/giac/spkg-src: Make the script work on macOS too

Changed commit from 31d2b93 to 84df0e5

Branch pushed to git repo; I updated commit sha1. New commits:

84df0e5build/pkgs/giac: New upstream tarball

Changed upstream from Reported upstream. No feedback yet. to Fixed upstream, in a later stable release.

Author: Matthias Koeppe

Reviewer: François Bissey

comment:22

LGTM. I didn't even know you could use that syntax for tar under linux, looks much better now. Have to check a couple of things before switching it to positive.

comment:23

does this work? I would worry about - in the patch level.

Changed reviewer from François Bissey to none

Reviewer: François Bissey, Dima Pasechnik

comment:25

Attachment: giac-1.5.0.63-p0.tar.bz2.gz

Branch pushed to git repo; I updated commit sha1. New commits:

081b1d2Fixup
a9b4666fixup

Changed commit from 84df0e5 to a9b4666

comment:26

This one works for me.

comment:27

Using -p0 because that becomes part of the tarball name.

comment:28

afk now, but OK.

comment:29

I checked the bit I was worried about (whether some stuff in doc/Makefile.am could lead to a breakage) and I am definitely OK with this.

comment:30

Thanks!