Stefano, many thanks.
This will be a hard challenge for me to obey to the rules :-)
The 80 character thing is reported from checkpatch.pl as WARNING, not as ERROR - so there will be a chance of the patch to become accepted by a custodian.
In between i tried to reformat by Code to obey to these rules, but now checkpatch.pl tells me following:
WARNING: Avoid unnecessary line continuations #531: FILE: board/BuR/bur_tseries/board.c:155: + if (tps65217_voltage_update(TPS65217_DEFDCDC3, \
in real life the code section looks like this: /* Set DCDC3 (CORE) voltage to 1.125V */ if (tps65217_voltage_update(TPS65217_DEFDCDC3, \ TPS65217_DCDC_VOLT_SEL_1125MV)) { puts("tps65217_voltage_update failure\n"); return; }
If i don't do the line break at line 155 checkpatch.pl is disturbing me with the 80 character rule.
Another thing is linewrapping of output strings, to obey to the rules i have to format the string as following:
if (i2c_probe(TPS65217_CHIP_PM)) { printf("PMIC chip (0x%02x) not present! skipping" \ "further configuration.\n", TPS65217_CHIP_PM); return; }
But this makes it impossible to grep the code in case of an error.
Any idea how to deal with such things ?
best regards, Hannes
Stefano Babic wrote:
Hi Hannes,
On 04/02/2014 13:50, Hannes Petermaier wrote:
Hello,
I've read on the U-Boot website about coding style. They say 'All contributions to U-Boot should conform to the Linux kernel coding style'. Further i've read this link and there are at least 2 things which i have troubles with.
a tab-ident is 8 spaces
Right
(in eclipse for example with the built in K&R has 4 characters). With 8 characters per TAB the problem from point 2 enters very, very fast.
a line ist limited to 80 characters (maybe thus comes from very old days where displays couldn't show more than this and scrolling was very expensive). is this state of the art ?
Yes, it is. Same rule as in kernel.
In fact i've found a lot of files within the u-boot code which do not obey to this rules.
I cannot tell for each exceptions you found. Maybe the code is very old, and it slipped into mainline without fixing the lenght. There are also some well-known exception (generally for tables), where a longer line was accepted after discussion in the ML to increase readability. (tables for pinmux, as example arch/arm/include/asm/arch-mx6/mx6dl_pins.h).
Now my question is, how strong are this two points ?
They are strong until there is a general acceptance on the ML to drop the rule. However, the rule was already discussed in the past (you can dig deep in archives), and the result was to maintain it, exactly as it is valid for linux.
checkpatch.pl rails against a patch file with a lot of warnings if there is a line longer than 80bytes, also it takes TABS as 8 spaces.
checkpatch is run before merging a patch into a merge tree by custodians. If checkpatch reports errors, they must be fixed.
whats your opinion about this ?
Last question: Is this mailing list the right place of discussing such things ?
I think you are in the right place to discuss these topics ;-)
Best regards, Stefano Babic
--
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: +49-8142-66989-53 Fax: +49-8142-66989-80 Email: sbabic@denx.de =====================================================================