failure to evolve a std::map value #20421
Description
Check duplicate issues.
- Checked for duplicates
Description
I have a type (called Header in the reproducer) that is the value type of a std::map stored in a TTree. I can read/write this branch header_map of the TTree just fine, but when I apply read schema rules to the Header class, all of the values are then 0 - seemingly like what was observed in #19650 - so I'm guessing that there is a similar address-setting issue.
$ ./write # writes ten entries into the TTree
$ ./read-v1 # reads with original version used to write
{ 0: {run: 42, event: 0} }
{ 1: {run: 42, event: 10} }
{ 2: {run: 42, event: 20} }
{ 3: {run: 42, event: 30} }
# etc...
$ ./read-v2 # reads with updated version with schema rules
{ 0: {applying v1->v2 schema evolution rule
run: 0, event: 0} }
{ 1: {applying v1->v2 schema evolution rule
run: 0, event: 0} }
{ 2: {applying v1->v2 schema evolution rule
run: 0, event: 0} }
{ 3: {applying v1->v2 schema evolution rule
run: 0, event: 0} }
# etc...
Reproducer
The reproducer is available on GitHub: https://github.com/tomeichlersmith/ldmx-root-schema-evolution-testbench/tree/main/forum/schema-evolve-std-map-value
(The forum/schema-evolve-std-map-value subdirectory of tomeichlersmith/ldmx-root-schema-evolution-testbench`.)
ROOT version
ROOT Version: 6.38.00-rc1
Platform: linuxx8664gcc
Compiler: 13.3.0
Specifically, I am using a container image where ROOT is configured and built as defined here
Installation method
build from source
Operating system
Linux (Ubuntu 24.04)
Additional context
Found while testing my solution to my forum post about schema evolution while slow cloning with more types that are in use in ldmx-sw. https://root-forum.cern.ch/t/quiet-writing-failure-when-applying-schema-evolution-while-slow-cloning/64350
In that forum post, I was looking at just using the current-LDMX-standard ROOT 6.34, but it has this issue so I updated to 6.38.00-rc1 to see if it had been resolved since.
Potentially Related
- https://root-forum.cern.ch/t/manual-schema-evolution-with-i-o-rules-and-branches-containing-vector-t/64026
- [tree] missing staged data in I/O rules of split vector<T> --> vector<U> #19650 -> resolved by [tree] fix TBranchElement for split collection with renamed inner type #19688 (included in 6.38.00-rc1)
- [tree] missing value in I/O rule for vector<T> with deep source member #19773 -> maybe also is the cause of this issue I'm observing?