3 ms·

Hmm, that's not very comprehensive for this purpose. Years ago at a former employer, we discovered just after shipping a large quantity of quite sensitive demo

Hmm, that's not very comprehensive for this purpose. Years ago at a former employer, we discovered just after shipping a large quantity of quite sensitive demo materials that an outside contractor had managed to slip a hidden profanity into it. Oh, the joy that caused.... The immediate reaction though was that a few of the less polite and proper members of the organisation were tasked with producing as close to an exhaustive profanity list as they could so we could do a relatively thorough sweep. From memory that list was pushing 40 terms - I think I might still have a copy somewhere but the last thing this discussion needs is more swearing!

barake 16y ago
We have a table full of unsavory words due to generating more than one password that upset someone. At one point someone decided we needed "pronounceable" passwords for our new users without considering the consequences.
- sbov 16y ago
  Random numbers can also you in trouble in this way. E.g. anything with 666. The hard part is cross culture references that you may be unaware of.
- eftpotrm 16y ago
  I had a brief play once for this sort of thing with doing stem-based pronounceable word generation based off a starting dictionary; the idea being that it broke them into fragments marked as beginning, middle or end and used intersections of these to build words that followed letter use rules of known words but yet weren't known words themselves. You start getting distance problems though - a lot of words it generates are too close to or are trivial encapsulations of existing words, so that'd need taking account of. One day I'll have a proper play and do something more comprehensive for this. Tip for anyone cacheing unsavoury words - from memory the bulk of what we ended up checking for were racial slurs, of which there's an astonishing (and unsurprisingly heavily localised) variety.
adriand 16y ago
Years ago I developed a custom CRM for a client that included reminders. While testing the production version of the system pre-launch, I added many fake reminders, for users who happened to be real, along with a lot of other test data. I then cleared all of the test data...except I accidentally skipped the reminders table. For weeks afterwards, my clients would receive profanity-laden reminders, helpfully labeled as "from Adrian". Luckily, they had a good sense of humour about it. But since then, I've never used profanity in any kind of test data.
- anthonyb 16y ago
  or at least not used your real name...
- dspillett 16y ago
  I never use anything remotely non-serious in my test data. I even try make comments type on forms in tested seem at least close to real. Silly names are out too. You never know when some fool is going to show a visiting client a development system instead of one of the official demos, and I've been stung by someone not being amused by the existence of Don Kiddick and Mike Hunt in my sample dataset.

barake 16y ago

We have a table full of unsavory words due to generating more than one password that upset someone. At one point someone decided we needed "pronounceable" passwords for our new users without considering the consequences.

sbov 16y ago
Random numbers can also you in trouble in this way. E.g. anything with 666. The hard part is cross culture references that you may be unaware of.
eftpotrm 16y ago
I had a brief play once for this sort of thing with doing stem-based pronounceable word generation based off a starting dictionary; the idea being that it broke them into fragments marked as beginning, middle or end and used intersections of these to build words that followed letter use rules of known words but yet weren't known words themselves. You start getting distance problems though - a lot of words it generates are too close to or are trivial encapsulations of existing words, so that'd need taking account of. One day I'll have a proper play and do something more comprehensive for this. Tip for anyone cacheing unsavoury words - from memory the bulk of what we ended up checking for were racial slurs, of which there's an astonishing (and unsurprisingly heavily localised) variety.

adriand 16y ago

Years ago I developed a custom CRM for a client that included reminders. While testing the production version of the system pre-launch, I added many fake reminders, for users who happened to be real, along with a lot of other test data. I then cleared all of the test data...except I accidentally skipped the reminders table. For weeks afterwards, my clients would receive profanity-laden reminders, helpfully labeled as "from Adrian". Luckily, they had a good sense of humour about it. But since then, I've never used profanity in any kind of test data.

anthonyb 16y ago
or at least not used your real name...
dspillett 16y ago
I never use anything remotely non-serious in my test data. I even try make comments type on forms in tested seem at least close to real. Silly names are out too. You never know when some fool is going to show a visiting client a development system instead of one of the official demos, and I've been stung by someone not being amused by the existence of Don Kiddick and Mike Hunt in my sample dataset.