Telnet: Write full buffer instead of byte-by-byte

[richardhsu]

Previous TODO wanting to write in chunks. We should support writing more
at once since some TELNET servers may respond immediately upon first
byte written such as WHOIS servers.

---

Was attempting to use CURL to connect to WHOIS servers and found that certain servers didn't work well with send_telnet_data sending byte-by-byte. Saw TODO so attempted to do it.

Have the following concerns:

• TODO states to do chunk by chunk, not sure what a standard practice may be for the sizing so for initial PR ended up making a full escaped out buffer and just let the Curl_write send as much as possible and loop until completion.
• Not sure how to write tests for this but ran existing tests which all passed and ended up testing a local C program to use CURL against godaddy's WHOIS server which doesn't work until this patch is applied. If need tests, any advice on that would be appreciated!

Thanks

[mention-bot]

@richardhsu, thanks for your PR! By analyzing the history of the files in this pull request, we identified @bagder, @yangtse and @dfandrich to be potential reviewers.

[bagder]

We don't have any TELNET test server so the only possible way right now, without adding such a test server, is probably to send a HTTP 1.0 request via TELNET to the HTTP test server and make sure that works as expected.

[bagder]

I think the change looks fine. The double looping over the contents is unfortunate, but I can't see an easy way to avoid that.

[richardhsu]

@bagder thanks! Also added a test1399 to do normal TELNET stdin to the HTTP test server :)

[richardhsu]

@bagder looks like checks and tests pass, let me know if I should add anything more in order to get this merged, thank you!

I've also been running this on my local machine for my use case and it is working as expected.

[bagder]

My gcc complains since there's a comparison of a signed character with 255 (CURL_IAC) that will never equal true:

telnet.c: In function ‘send_telnet_data’:
telnet.c:1232:18: error: comparison is always false due to limited range of data type [-Werror=type-limits]
     if(buffer[i] == CURL_IAC)
                  ^~
telnet.c:1246:20: error: comparison is always false due to limited range of data type [-Werror=type-limits]
       if(buffer[i] == CURL_IAC)
                    ^~
In file included from telnet.c:59:0:
arpa_telnet.h:77:19: error: overflow in implicit constant conversion [-Werror=overflow]
 #define CURL_IAC  255 /* Interpret As Command */
                   ^
telnet.c:1247:23: note: in expansion of macro ‘CURL_IAC’
         outbuf[j++] = CURL_IAC;
                       ^~~~~~~~

[richardhsu]

@bagder Ah, yes, thanks, updated the outbuf to be unsigned char * and then added casts during comparisons between buffer and the various unsigned chars.

[richardhsu]

@bagder Hi, was wondering if there's any other issues before getting this merged? Thanks :)

[bagder]

I still see two compiler warnings/errors (./configure --enable-debug adds picky options):

telnet.c: In function ‘send_telnet_data’:
telnet.c:1237:12: error: pointer targets in assignment differ in signedness [-Werror=pointer-sign]
     outbuf = buffer;
            ^
telnet.c:1246:20: error: comparison is always false due to limited range of data type [-Werror=type-limits]
       if(buffer[i] == CURL_IAC)
                    ^~

[richardhsu]

@bagder Ah awesome, thanks, the --enable-debug was helpful, missed the two other casts needed but have now corrected them. Thanks!

[bagder]

thanks!

[MarcelRaad]

The new test hangs infinitely on MinGW, also all the MinGW autobuilds have been failing since then. I haven't found time to debug this yet.

[richardhsu]

Oh hmm, sorry about that. Looks like 9e093f0#diff-7b143a8d3555776816bc29d9cb83e6a2 actually addresses such issue I think. I believe then the test1399 added isn't needed since test1326 tests TELNET behavior which that test already passes since the suite has been passing.