Re: [boost] CVS main line is all messed up

From: Beman Dawes (bdawes_at_[hidden])
Date: 2003-03-16 16:06:55

At 09:30 AM 3/14/2003, Victor A. Wagner, Jr. wrote:

 >I appreciate the difficulties in getting a release out.
 >I _am_ puzzled about what behavior you wish people (who use (all | any) 
of
 >boost regularly) to use in validating problems they may encounter (most 
of
 >the time boost is in a constant state of improvement).
 >During these times, the usual advice is check the latest out of CVS and
 >verify the problem exists there.

No, the intent wasn't for people to do anything different..

 >Now we appear to be told that the "latest" isn't to be used (for 
filesystem
 >at any rate) until 1.30 is released.
 >Is this a rule for ALL of boost for the duration?
 >Enquiring minds want to know.
 >
 >_Surely_ you don't want people to _quit_ testing during this pre-release
 >phase.  That would make the whole phase irrelevant.
 >
 >I suggest further, that perhaps the release mechanism be changed such 
that
 >the "how to check the latest" NEVER changes from the point of view of the 

 >user/tester i.e. "cvs update -A -P -d" would ALWAYS get the latest 
believed
 >to be working copy.

During release preparation, there are two actively being worked on version 
- the main trunk and the RC branch.

One isn't better than the other - they are just different.

The reason I didn't commit the VC++7.1 beta filesystem operators_test 
workaround to the main trunk and then merge right away into the release 
candidate was that I simply don't know at this point how we are going to 
handle the new bugs in VC+7.1. It hasn't even been shipped yet, and won't 
for another month or so.

Please don't assume that this sort of very specific workaround is 
indicative of policy - it is just a specific compiler problem workaround 
for a specific release.

--Beman