| 03:12.35 | *** join/#stepcode hoiji (~hoiji@117.201.180.164) |
| 07:52.41 | *** join/#stepcode starseeker (~starseeke@66-118-151-70.static.sagonet.net) |
| 07:52.41 | *** join/#stepcode brlcad (~sean@66-118-151-70.static.sagonet.net) |
| 09:50.33 | *** join/#stepcode biochem (~biochem@195.88.191.131) |
| 10:18.17 | *** join/#stepcode hoiji (~hoiji@117.201.87.137) |
| 11:00.49 | *** join/#stepcode hoiji (~hoiji@117.208.68.171) |
| 11:39.00 | *** join/#stepcode hoiji (~hoiji@117.201.81.186) |
| 12:03.21 | *** join/#stepcode hoiji (~hoiji@59.89.18.68) |
| 12:42.42 | *** join/#stepcode blackmac (~blackmac@82.146.41.49) |
| 14:17.01 | *** join/#stepcode cwstirk (~charlie@c-71-56-216-45.hsd1.co.comcast.net) |
| 14:39.07 | *** join/#stepcode hoiji (~hoiji@117.201.92.188) |
| 17:15.49 | hoiji | Hi everyone I am working towards making a gsoc proposal to make STEP Libraries thread safe. Many of the data structures being used in the libraries take dependency on src/base/judy*. Can I assume the judy array implementations in particular to be thread safe. Or should I investigate them also? |
| 17:41.58 | mpictor | hoiji: I doubt that it is thread safe |
| 18:01.30 | hoiji | Ok, Would it be alright to put mutexes around access to the judy data-structure which I think are critical,earlier in the project. Once I am done with making the class libraries thread safe, I will try moving the mutexes inside the judy array implementation to make the critical regions smaller. |
| 18:08.43 | brlcad | hoiji: glad to hear the interest |
| 18:09.25 | brlcad | you should confirm whether it is thread safe or not before wrapping it in mutexes, no? |
| 18:22.48 | hoiji | brlcad: What you said is true. |
| 18:24.00 | hoiji | What I am thinking of is of top-down approach, i.e assume the worst case scenario that judy implementation is not thread safe. Then make the src/cl* libraries thread safe. After that I can dwelve into judy implementation and find whther they are thread safe or not, and depending on the findings remove/move mutexes if necessary. |
| 18:24.51 | hoiji | Of-course I can go the other way around (bottom to top). By first finding about the thread safety of judy approach. And then making the higher level functions thread safe. |
| 18:25.00 | hoiji | Both approaches have their pros and cons |
| 18:25.23 | hoiji | But since I am stranger to most of the source code (except lazy loading). It would be better for me to go with what experienced developers like you say :) |
| 18:39.15 | brlcad | hoiji: I'd leave that entirely for mpictor to decide |
| 18:39.19 | brlcad | but to me adding mutexes before figuring out they're required breaks a concept of coding complete |
| 18:39.59 | brlcad | basically, if you were hit by a bus or won the lotter and were never seen from again mid-project, did you leave turds that someone else would have to clean up or investigate? |
| 18:40.10 | brlcad | s/lotter/lottery/ |
| 18:41.14 | brlcad | infobot: chanset #stepcode sed 0 |
| 18:47.51 | hoiji | brlcad: Thanks for the analogy, I do understand the folly in my idea. It was probably because its the first time I am working in an open source project. |
| 18:48.08 | hoiji | But to make things clear even if I win a lottery I wont abondon the project mid-way. I hope I am not hit by a bus though |
| 19:03.57 | brlcad | maybe both! |
| 19:04.51 | brlcad | hit be a bus on your way to pick up lottery winnings, or hit by a lottery bus (you win!) |
| 19:46.25 | *** join/#stepcode starbaths (~starbaths@91.207.117.35) |
| 21:07.33 | mpictor | would prefer a very near miss with a lottery bus ;) |
| 21:38.50 | *** join/#stepcode newzeus (~newzeus@82.146.61.35) |
| 22:53.54 | *** join/#stepcode mrivcrealty (~mrivcreal@188.190.126.43) |