MantisBT - CMake |
| View Issue Details |
|
| ID | Project | Category | View Status | Date Submitted | Last Update |
| 0013106 | CMake | CMake | public | 2012-04-06 05:03 | 2013-01-09 14:05 |
|
| Reporter | tfischer | |
| Assigned To | David Cole | |
| Priority | normal | Severity | minor | Reproducibility | random |
| Status | closed | Resolution | no change required | |
| Platform | Gentoo Linux | OS | Linux | OS Version | |
| Product Version | CMake 2.8.6 | |
| Target Version | CMake 2.8.10 | Fixed in Version | CMake 2.8.10 | |
|
| Summary | 0013106: SystemTools::CopyFileAlways fails at random due to invalid kwsys_ios::ifstream fin |
| Description | I am using a custom cmake script to determine the SVN revision number of my source code during build time and to write this version number into a file "version.h" so that the compiler can include the version string into the compiled code.
The script works as follows: First, it determines the SVN revision (svnversion or environment variable) and writes a small header file called version.h.tmp. To avoid unnecessary recompilations, cmake's copy_if_different is used to copy version.h.tmp to version.h only if the SVN revision (therefore the file's content) has changed.
When testing the script, in irregular intervals the copy_if_different call fails with error message:
Error copying file (if different) from "/home/tf/programming/gna/qtcreator-build/src/parts/version.h.tmp" to "/home/tf/programming/gna/qtcreator-build/src/parts/version.h".
I added print statements before each "return false" statement in SystemTools::CopyFileAlways (eventually called by copy_if_different) to check which test fails. As it turns out, the test for !fin fails in line 1849 (for CMake 2.8.6).
The problem now is, that this failure does not follow any pattern as far as I can see. Running my custom script (attached to this report) 20 times in a loop (bash for-loop), the following output is generated.
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
kwsys_ios::ifstream fin is NULL
Error copying file (if different) from "/home/tf/programming/gna/qtcreator-build/src/parts/version.h.tmp" to "/home/tf/programming/gna/qtcreator-build/src/parts/version.h".
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
kwsys_ios::ifstream fin is NULL
Error copying file (if different) from "/home/tf/programming/gna/qtcreator-build/src/parts/version.h.tmp" to "/home/tf/programming/gna/qtcreator-build/src/parts/version.h".
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
-- SVN version was set outside through environment variable SVN_REV
-- SVN version is AAA
kwsys_ios::ifstream fin is NULL
Error copying file (if different) from "/home/tf/programming/gna/qtcreator-build/src/parts/version.h.tmp" to "/home/tf/programming/gna/qtcreator-build/src/parts/version.h".
The for-loop to call this script is:
for n in {1..20} ; do /usr/bin/cmake -DSOURCE_DIR=/home/tf/programming/gna/proj/src/parts -DBINARY_DIR=/home/tf/programming/gna/qtcreator-build/src/parts -P /home/tf/programming/gna/proj/src/getsvn.cmake ; done
Any ideas what could cause this behaviour? How to debug/trace this problem?
|
| Steps To Reproduce | |
| Additional Information | |
| Tags | No tags attached. |
| Relationships | |
| Attached Files |  getsvn.cmake (2,630) 2012-04-06 05:03
https://public.kitware.com/Bug/file/4289/getsvn.cmake
race.cmake (815) 2012-08-22 16:11
https://public.kitware.com/Bug/file/4432/race.cmake |
|
| Issue History |
| Date Modified | Username | Field | Change |
| 2012-04-06 05:03 | tfischer | New Issue | |
| 2012-04-06 05:03 | tfischer | File Added: getsvn.cmake | |
| 2012-08-11 21:35 | David Cole | Status | new => backlog |
| 2012-08-11 21:35 | David Cole | Note Added: 0030389 | |
| 2012-08-22 16:11 | tfischer | Note Added: 0030738 | |
| 2012-08-22 16:11 | tfischer | File Added: race.cmake | |
| 2012-08-22 16:27 | David Cole | Assigned To | => David Cole |
| 2012-08-22 16:27 | David Cole | Status | backlog => assigned |
| 2012-08-22 16:32 | David Cole | Note Added: 0030741 | |
| 2012-08-22 16:32 | David Cole | Status | assigned => resolved |
| 2012-08-22 16:32 | David Cole | Fixed in Version | => CMake 2.8.10 |
| 2012-08-22 16:32 | David Cole | Resolution | open => no change required |
| 2012-08-22 16:32 | David Cole | Target Version | => CMake 2.8.10 |
| 2013-01-09 14:05 | Robert Maynard | Note Added: 0032079 | |
| 2013-01-09 14:05 | Robert Maynard | Status | resolved => closed |
|
Notes |
|
|
(0030389)
|
|
David Cole
|
|
2012-08-11 21:35
|
|
Sending old, never assigned issues to the backlog.
(The age of the bug, plus the fact that it's never been assigned to anyone means that nobody is actively working on it...)
If an issue you care about is sent to the backlog when you feel it should have been addressed in a different manner, please bring it up on the CMake mailing list for discussion. Sign up for the mailing list here, if you're not already on it: http://www.cmake.org/mailman/listinfo/cmake [^]
It's easy to re-activate a bug here if you can find a CMake developer who has the bandwidth to take it on, and ferry a fix through to our 'next' branch for dashboard testing.
|
|
|
|
(0030738)
|
|
tfischer
|
|
2012-08-22 16:11
|
|
I was able to trace this problem further. It seems that it makes a difference if "copy_if_different" and "remove" are executed in the same "execute_process" or in two different "execute_process" statements.
I haven't checked that in the sources, but my guess would be that the two command statements inside the execute_process statement are executed in parallel (resulting in a race condition as observed) contrary to the documentation which states "Runs the given sequence of one or more commands ..." (emphasis on "sequence", i.e. sequential, which is not parallel).
Please have a look on the attached example script and try to confirm my observation. |
|
|
|
(0030741)
|
|
David Cole
|
|
2012-08-22 16:32
|
|
You are observing behavior that is by design. Sorry nobody took a closer look at this issue and got back to you before now.
The documentation for execute_process clearly states "Runs the given sequence of one or more commands with the standard output of each process piped to the standard input of the next."
The standard output of cmd1 is *PIPED* into the stdin of cmd2, so of course they're both going to be running at the same time.
It's as if you did this in a command prompt:
cmake -E copy_if_different f1 f2 | cmake -E remove f1
You are probably thinking that it's as if you used "&&" instead of "|" in the command prompt...
So... your solution is easy. Please use two separate execute_process commands.
Perhaps the documentation should be rephrased to avoid the word sequence, since the commands will run simultaneously as they produce and consume output/input. But other than using the word "sequence" in the documentation, there's no issue here. This is behaving the way it is intended. |
|
|
|
(0032079)
|
|
Robert Maynard
|
|
2013-01-09 14:05
|
|
|
Closing resolved issues that have not been updated in more than 4 months. |
|