[1.33.0] Feature freeze reminder

We are in feature freeze. No new features can be added to any library without specific approval from me. We still have many, many bugs to fix before this release. We CAN NOT afford to create any more bugs. Please analyze and fix the bugs you have. Doug Gregor 1.33.0 Release Manager

Hi Doug, "Doug Gregor" <dgregor@cs.indiana.edu> wrote in message news:74c13f31bb7a782cd1fdd42ed76ac3ee@cs.indiana.edu... | We are in feature freeze. No new features can be added to any library | without specific approval from me. We still have many, many bugs to fix | before this release. We CAN NOT afford to create any more bugs. Please | analyze and fix the bugs you have. I've updated assign/range/ptr_container (probably) for the last time. The dmc compiler doesn't seem to have any configuration yet, so it is doomed to fail most tests; I've simply mark it as unusable for now. I have one serious error left in ptr_list for gcc 4.0.0 which I will try to get some help fixing. -Thorsten

On May 20, 2005, at 6:46 PM, Thorsten Ottosen wrote:
Hi Doug,
"Doug Gregor" <dgregor@cs.indiana.edu> wrote in message news:74c13f31bb7a782cd1fdd42ed76ac3ee@cs.indiana.edu... | We are in feature freeze. No new features can be added to any library | without specific approval from me. We still have many, many bugs to fix | before this release. We CAN NOT afford to create any more bugs. Please | analyze and fix the bugs you have.
I've updated assign/range/ptr_container (probably) for the last time. The dmc compiler doesn't seem to have any configuration yet, so it is doomed to fail most tests; I've simply mark it as unusable for now.
No need to worry about that compiler. The compilers listed in the "release" mode are highest priority; dmc we're not even going to look at until after the release.
I have one serious error left in ptr_list for gcc 4.0.0 which I will try to get some help fixing.
Okay, thanks. Doug

I have one more thing I want to address. I've checked in shared_ptr serialization which uses a new approach. This reconciles different view points on this topic. The new version passes all tests on most compilers. The problem is that I want to add code that will handle the previous pointer version. My concern is that users might want to use 1.33 to read archives created under 1.32 and without this addition this won't be possible. I presume that some numbers of people would be effected by such a situation. If not I could just forget it. Robert Ramey Doug Gregor wrote:
We are in feature freeze. No new features can be added to any library without specific approval from me. We still have many, many bugs to fix before this release. We CAN NOT afford to create any more bugs. Please analyze and fix the bugs you have.
Doug Gregor 1.33.0 Release Manager
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

The problem is that I want to add code that will handle the previous pointer version. My concern is that users might want to use 1.33 to read archives created under 1.32 and without this addition this won't be possible. I presume that some numbers of people would be effected by such a situation. If not I could just forget it.
Without backwards compatibility with the old shared pointer serialization I for one would not be able to upgrade to the latest version until it appears. All a pain really as I was looking forward to the new improved version. cheers Martin -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.322 / Virus Database: 266.11.14 - Release Date: 20/05/2005

"Martin Slater" <mslater@hellinc.net> wrote in message news:428EBB2F.5060101@hellinc.net...
The problem is that I want to add code that will handle the previous pointer version. My concern is that users might want to use 1.33 to read archives created under 1.32 and without this addition this won't be possible. I presume that some numbers of people would be effected by such a situation. If not I could just forget it.
Without backwards compatibility with the old shared pointer serialization I for one would not be able to upgrade to the latest version until it appears. All a pain really as I was looking forward to the new improved version.
I'll second that. Jeff Flinn

Robert Ramey wrote:
The problem is that I want to add code that will handle the previous pointer version. My concern is that users might want to use 1.33 to read archives created under 1.32 and without this addition this won't be possible. I presume that some numbers of people would be effected by such a situation. If not I could just forget it.
We definitely need this too. Our default data file format is a xml_archive, which contains many shared_ptr's. Backwards compatibility is very important to us and if it isn't provided in 1.33, we won't be upgrading. Thanks Russell

The problem is that I want to add code that will handle the previous pointer version. My concern is that users might want to use 1.33 to read archives created under 1.32 and without this addition this won't be possible. I presume that some numbers of people would be effected by such a situation. If not I could just forget it.
I would guess that most people would view the ability to read archives created with 1.32 as more or less vital, just my 2c worth... I don't really see fixing this as an addition, more like fixing a regression, as long is doesn't hold up the release of course! If the latter is likely, then delay until after the release, and then release a patch against 1.33: maybe put a note in the 1.33 docs that a patch is imminent, but didn't make the deadline etc. John.

On May 20, 2005, at 8:49 PM, Robert Ramey wrote:
I have one more thing I want to address.
I've checked in shared_ptr serialization which uses a new approach. This reconciles different view points on this topic. The new version passes all tests on most compilers.
The problem is that I want to add code that will handle the previous pointer version. My concern is that users might want to use 1.33 to read archives created under 1.32 and without this addition this won't be possible. I presume that some numbers of people would be effected by such a situation. If not I could just forget it.
As others have also said, breaking compatibility with archives created by 1.32 should be considered a bug (that will affect a lot of users). Please go ahead and make these changes. Doug

We are in feature freeze. No new features can be added to any library without specific approval from me. We still have many, many bugs to fix before this release. We CAN NOT afford to create any more bugs. Please analyze and fix the bugs you have.
Config, type traits, static assert, and regex are frozen for release now (and I'm away next week anyway). Regards, John.
participants (8)
-
Doug Gregor
-
Douglas Gregor
-
Jeff Flinn
-
John Maddock
-
Martin Slater
-
Robert Ramey
-
Russell Hind
-
Thorsten Ottosen