Control: tags -1 -moreinfo

Hi Paul,

On Tue, Jun 18, 2019 at 7:23 PM Paul Gevers <elb...@debian.org> wrote:
> There are quite a lot of changes in the diff, and only at the end we get
> to the Reiwa support (if I understand the code and comments at all). You
> haven't explained these other changes. If all we are now talking about
> is Reiwa support I will *not* unblock the current version of icu, as it
> introduces changes that are not targeted at fixing that.
 Indeed, there are other changes. Let me go into more details via the
upstream commit list[1] which is on the maint-63 branch (important
bugfixes only for the 63 release).
The 63.1 release happened on Oct 13, 2018 with the last commit of
ICU-20205 (RelativeDateTimeFormatter pt data fix, improved error
handing and test)[2].
The first fix is ICU-20208 (uspoof.cpp function checkImpl should be
static) [3] and I've backported it with ICU-20246 (Fixing another
integer overflow in number parsing) [4] (security issues) to (and
already in) Buster.
Fixes between these commits (ICU-20214 Fix namespace error on Cygwin,
ICU-20209 Fix build failures on Windows and ICU-20240 Fix Windows
build failure) [5] are for Windows only and don't affect us.
Next is a faster operation change with hashtables (ICU-20250 faster
MutableCodePointTrie.build()) [6]. An additional speed change is the
startup regression change (ICU-20250 make UnicodeSet(intprop=value)
faster) [7] that I had to reverse apply (meaning not part of the
source in this unblock request) due to its ABI break.
The next commit is for the Java ICU source [8] (ICU is split in Debian
to src:icu and src:icu4j), ICU-20255 revert to reflection for methods
not yet in Android API thus not part of this unblock request.
Then comes the Japanese era Reiwa support commits [9] (ICU-20536),
these would be very good to have for Buster. Then the version number
update with the IANA tzdata update to 2019a [10] come. The latter is
preprocessed to the source/data/in/icudt63l.dat file.

In general the other changes you see are Windows only related changes.
What I didn't mention is the hashtable speedup fix[6] and the IANA
tzdata 2019a update which is preprocessed to the dat file.
I know there's one more commit [11] (ICU-20558 Fix regression in
DateTimePatternGenerator) but as I haven't seen it before it's also
not in this unblock request (ie, not yet backported). How this should
be handled? Do you need more information for icu-63.2-2 and/or can I
use this fix as well and need to ask you only then for the unblock?

Thanks for consideration,
Laszlo/GCS
[1]  https://github.com/unicode-org/icu/commits/maint/maint-63
[2]  
https://github.com/unicode-org/icu/commit/46895456ad1b6660d17eaeba2c101600ad8d8eb8
[3]  
https://github.com/unicode-org/icu/commit/8baff8f03e07d8e02304d0c888d0bb21ad2eeb01
[4]  
https://github.com/unicode-org/icu/commit/6cbd62e59e30f73b444be89ea71fd74275ac53a4
[5]  
https://github.com/unicode-org/icu/commit/78d61f7cb9ecd98664d134da5909c8cc3dcee5a6
     
https://github.com/unicode-org/icu/commit/fced46c6a9404719482a27a29baa371809a64f37
     
https://github.com/unicode-org/icu/commit/2094a48c20193d8ecee70dc39a18f61bfda41f25
[6]  
https://github.com/unicode-org/icu/commit/38b882fe30e7105e691c0d48d35864724ea72979
[7]  
https://github.com/unicode-org/icu/commit/f3fa0d604ef6527a01dab96f4bfa3c5290127337
[8]  
https://github.com/unicode-org/icu/commit/0af507c4dabe517d9daa621d2d8c59c74b04af59
[9]  
https://github.com/unicode-org/icu/commit/838a14ee42336270260b54e11cd453a94b3c1a2d
     
https://github.com/unicode-org/icu/commit/c948a1ae3132ab7e5af28f3321905dbaf97b0ce4
[10] 
https://github.com/unicode-org/icu/commit/4f715ae124c418a15a3fa4d8fb14f406576a7ee5
[11] 
https://github.com/unicode-org/icu/commit/5df4d7dfd8d77dd16aa3a0b398d50a22f4c85daa

Reply via email to