1

Topic: Boost and regression-tests

Greetings! Somebody tried to launch regresses-tests Boost? At them on a site it is told: Running Boost Regression Tests http://www.boost.org/development/runnin … tests.html To run a library's regression tests, run Boost's b2 utility from the <boost-root>/libs / <library>/test directory. To run a single test, specify its name (as found in <boost-root>/libs/<library>/test/Jamfile.v2) on the command line. Installed a Visual Studio 2015, delivered Update 3. Unpacked Boost 1.66 in a folder on a hard disk. Opened "Developer Command Prompt for VS2015" (console), I come in any , for example libs\smart_ptr\test and I launch the test a command b2. Not clearly that the considerable part of tests is filled up. For example, at me are not tested for libraries accumulators (problems with weighted_median and p_square_cumul_dist), assign, bimap, bind, date_time, geometry, graph, lockfree, phoenix... Did not look Further. Start b2 from Boost\status with different options in essence does not change anything. Imho, for library of such range somehow  . What is it? Such "quality" of tests Boost? Or most Boost? Or I do something incorrectly?

2

Re: Boost and regression-tests

Hello, okman, you wrote: O> Somebody tried to launch regresses-tests Boost? Boost' results of running: http://www.boost.org/development/tests/ … ators.html

3

Re: Boost and regression-tests

Hello, Voivoid, you wrote: V> Boost' results of running: the Photoshop!

4

Re: Boost and regression-tests

Hello, Voivoid, you wrote: V> Boost' results of running: V> http://www.boost.org/development/tests/ … ators.html Thanks, these labels I saw. And there tests for a branch master from any not clear audits git are resulted. And to release 1.66 there corresponds audit b143a5... (Looked in git showlog) - it in the table is not specified absolutely not. Another interests: how to receive "green strips" at itself it is local, i.e. to be convinced, what leaking  version Boost (1.66) really transits all tests on VS 2015 update 3? At me, for example, tests Boost. Bind under x64 do not transit, of what I do an output that the library is broken and to use to it it is impossible. I try to migrate from ancient version Boost on new, in passing skipping through some versions of the compiler. It would be desirable to secure itself against errors in another's code. One or the other: or I incorrectly launch tests (and how then it is correct?) Or "ideology" of tests in Boost is broken for a long time and to trust them generally it is impossible. By the way, checked up for the sake of interest on Boost 1.42.0 in MSVC-9.0 - there the same trouble, from the several arbitrarily taken libraries at once were tumbled down  on tests. In general, I will be grateful for any kick in the correct direction.

5

Re: Boost and regression-tests

O> Installed a Visual Studio 2015, delivered Update 3. Unpacked Boost 1.66 in a folder on a hard disk. Can it is necessary collect it, instead of unpack? And is better pass to Version 1.*.0 version 1.65.1 can to appear less stable.

6

Re: Boost and regression-tests

Hello, TimurSPB, you wrote: O>> Installed a Visual Studio 2015, delivered Update 3. Unpacked Boost 1.66 in a folder on a hard disk. TSP> can it is necessary collect it, instead of unpack? Collected Boost in all combinations (i.e. debug/release, static/shared, 32/64) - changed nothing. How much I understood, assembly Boost generally does not influence tests in any way, libraries during tests all the same anew gather in bin.v2.

7

Re: Boost and regression-tests

O> Collected Boost in all combinations (i.e. debug/release, static/shared, 32/64) - changed nothing. O> how much I understood, assembly Boost generally does not influence tests in any way, libraries during tests O> all the same anew gather in bin.v2. It is possible to try still 1.65.1 if there is no that that very necessary that is present only at the last version. Because in 1.66.0 not all is smooth, judging by published results of tests. http://www.boost.org/development/tests/ … d-mlt.html