StartupRetroshare() sometimes calls exit() instead of giving a useful error

[public]

There are many error situations which result in StartupRetroshare() calling exit(1). This makes providing a robust login/profile setup facilitity complex as I all the error conditions must be checked ahead of time to reduce the possibility for unrecoverable errors and the demon exiting. Alternately I have to run a sort of fork()'d process monitor thing (with all the IPC overhead that entails) and interpret the stderr messages in a way that is useful for the end user.

A quick grep of the source suggests that exit(1) might be quite a common error path :(

[csoler]

that is mainly because

• we don't use exception handling since it's not reliable on windows (apparently! At least drbob
  prefers not to use them)
• some errors should not even happen. The call to exit is normally the very last resort, when
  nothing else can be done. For instance if the RS config directory cannot be read nor created, etc.

If you need to call startupRetroShare(), you can add error codes as a return parameter and remove
the exits it contains. The errors can be handled above (in main()?)

On 12/05/2011 05:07 PM, Alex Stapleton wrote:

There are many error situations which result in StartupRetroshare() calling exit(1). This makes providing a robust login/profile setup facilitity complex as I all the error conditions must be checked ahead of time to reduce the possibility for unrecoverable errors and the demon exiting. Alternately I have to run a sort of fork()'d process monitor thing (with all the IPC overhead that entails) and interpret the stderr messages in a way that is useful for the end user.

A quick grep of the source suggests that exit(1) might be quite a common error path :(

---

Reply to this email directly or view it on GitHub:
#5

[public]

I'll look into making StartupRetroshare not do exit(1) but somehow I suspect that it will need some work to get it to leave everything in a sensible state so that calling it multiple times won't confuse anything.

[csoler]

totally. It's a bit like a microsoft windows starting routine ;-) I can look into it tonight if you
prefer.

On 12/06/2011 10:26 AM, Alex Stapleton wrote:

I'll look into making StartupRetroshare not do exit(1) but somehow I suspect that it will need some work to get it to leave everything in a sensible state so that calling it multiple times won't confuse anything.

---

Reply to this email directly or view it on GitHub:
#5 (comment)

[csoler]

I replaced the exit() by returns in rsinit.cc. I havn't tested all failure possibilities though, but
it's not a big change either.
The clean way to handle this would be exception handling, but we can't use it on windows, if I
remember correctly.

Tell me if it's ok now.

Cyril

On 12/06/2011 10:26 AM, Alex Stapleton wrote:

I'll look into making StartupRetroshare not do exit(1) but somehow I suspect that it will need some work to get it to leave everything in a sensible state so that calling it multiple times won't confuse anything.

---

Reply to this email directly or view it on GitHub:
#5 (comment)