Testers coding bug fixes directly?
-
So I'm a dev manager and our QC lead who is gaining proficiency in coding, though is by no means even a junior programmer has taken it upon himself to directly fix some easy bugs. This is certainly a faster way to get things fixed as our dev resources are severely limited. So far it's restricted to typos/grammar mistakes in hard-coded string UI messages and that type of thing. Good or bad? Thoughts?
Everyone's responses against so far have been a bunch of "What Ifs" (or clever cliches in Latin). What if allowing the QA to fix a typo brings the project in a day early, and the Dev Manager gets a nice huge bonus? What if a developer changes something he/she really doesn't understand and delays the project for weeks? Separation of responsibility is a nice guideline, but exceptions almost always prove the rule. Who's QA'ing the QA? Hopefully, the dev manager is. You may have guessed that I am a dev manager. "I am rarely happier than when spending entire day programming my computer to perform automatically a task that it would otherwise take me a good ten seconds to do by hand." - Douglas Adams
-
Ah, I did say "outside of variables" in my post. Changing variables is definitely inviting trouble. I agree with you on that point.
That is just it. The poster said they were changing variables, albeit hard coded ones.
The report of my death was an exaggeration - Mark Twain
Simply Elegant Designs JimmyRopes Designs
I'm on-line therefore I am. JimmyRopes -
Everyone's responses against so far have been a bunch of "What Ifs" (or clever cliches in Latin). What if allowing the QA to fix a typo brings the project in a day early, and the Dev Manager gets a nice huge bonus? What if a developer changes something he/she really doesn't understand and delays the project for weeks? Separation of responsibility is a nice guideline, but exceptions almost always prove the rule. Who's QA'ing the QA? Hopefully, the dev manager is. You may have guessed that I am a dev manager. "I am rarely happier than when spending entire day programming my computer to perform automatically a task that it would otherwise take me a good ten seconds to do by hand." - Douglas Adams
Michael Haines wrote:
You may have guessed that I am a dev manager.
Then you are asking for trouble. Any changes "out of band" should not be done. There is a process in place to insure that casual changes do not take place. It is there for a very good reason. Regardless of all good intentions people will make mistakes. Sorry if your bonus is not as big, but the process has to be followed if consistency is to be maintained.
The report of my death was an exaggeration - Mark Twain
Simply Elegant Designs JimmyRopes Designs
I'm on-line therefore I am. JimmyRopes -
RyanDev wrote:
Heck, in one of my jobs we didn't even have QA. When I was doing asp we did the changes right on the production server sometimes.
I worked in a place like that. What a mess. :suss:
The report of my death was an exaggeration - Mark Twain
Simply Elegant Designs JimmyRopes Designs
I'm on-line therefore I am. JimmyRopes -
So I'm a dev manager and our QC lead who is gaining proficiency in coding, though is by no means even a junior programmer has taken it upon himself to directly fix some easy bugs. This is certainly a faster way to get things fixed as our dev resources are severely limited. So far it's restricted to typos/grammar mistakes in hard-coded string UI messages and that type of thing. Good or bad? Thoughts?
As long as it's limited to typo level fixes I don't have a problem with it, provided you have someone reviewing every change the tester makes. I'd be leery of letting someone just brought in to test try to fix anything more complex though; for the same reason I'd be leery about bringing a new dev into the project a week before release. I'm not as dogmatic as Mark is, because real world staffing constraints mean you can't always do things the best way; and it's still a step up from the same group of devs who wrote the app doing all the testing. (Nevermind not actually doing any formal testing before release at all. *shudder* BTDTGTWTF)
Did you ever see history portrayed as an old man with a wise brow and pulseless heart, waging all things in the balance of reason? Is not rather the genius of history like an eternal, imploring maiden, full of fire, with a burning heart and flaming soul, humanly warm and humanly beautiful? --Zachris Topelius Training a telescope on one’s own belly button will only reveal lint. You like that? You go right on staring at it. I prefer looking at galaxies. -- Sarah Hoyt
-
It's both good and bad for all the reasons already stated. There might be a reason it was coded a certain way that he doesn't know about, and the bug is actually somewhere else. So even if he fixed the problem he encountered he might have created another bug instead. So make sure his fixes gets sent back to the dev team for review.
Wrong is evil and must be defeated. - Jeff Ello[^]
Even better the QA guy can tell the place in code that he thinks should be fixed. That way devs don't waste time looking for right method but can spot if the QA guy was wrong and is just going to mask some other bug.
-- "My software never has bugs. It just develops random features."
-
It amuses me to read about the coder and QA tester being different people. You people don't know how good you have it... ;P
Heh, I used to work for company that looked at QA as money waste. You know - they don't "produce" anything and slow down development as instead of new features they want the bugs fixed... It was quite a few years ago. I'm not sure if they changed their mind or just died in pain :)
-- "My software never has bugs. It just develops random features."
-
Everyone's responses against so far have been a bunch of "What Ifs" (or clever cliches in Latin). What if allowing the QA to fix a typo brings the project in a day early, and the Dev Manager gets a nice huge bonus? What if a developer changes something he/she really doesn't understand and delays the project for weeks? Separation of responsibility is a nice guideline, but exceptions almost always prove the rule. Who's QA'ing the QA? Hopefully, the dev manager is. You may have guessed that I am a dev manager. "I am rarely happier than when spending entire day programming my computer to perform automatically a task that it would otherwise take me a good ten seconds to do by hand." - Douglas Adams
I'm sorry but it won't be faster. If your team is small then your devs probably know the app inside out. Let the QA guy tell the devs where he thinks the change should be made. If he is correct, it will be 2 mins to get it fixed. If there is something else - the hardcoded variable is used elsewhere the QA guy doesn't know about - the dev will know it and fix it properly. Changes outside of the dev team - especially small one - will make things much slower in future and harder for maintenance. I know every manager wants to deliver on time and the pesky QA guys just go in the way ;) I think it is better to ship with small visible bugs the client knows about (QA reported, but no time to fix), than ship with masked bugs as they will hit you harder later.
-- "My software never has bugs. It just develops random features."
-
So I'm a dev manager and our QC lead who is gaining proficiency in coding, though is by no means even a junior programmer has taken it upon himself to directly fix some easy bugs. This is certainly a faster way to get things fixed as our dev resources are severely limited. So far it's restricted to typos/grammar mistakes in hard-coded string UI messages and that type of thing. Good or bad? Thoughts?
Enforce Code Review process before any check-ins and then you know whether the fix he made is good or bad. Secondly the bugs that he fixes can be assigned to another tester for second round of testing. :java:
-
So I'm a dev manager and our QC lead who is gaining proficiency in coding, though is by no means even a junior programmer has taken it upon himself to directly fix some easy bugs. This is certainly a faster way to get things fixed as our dev resources are severely limited. So far it's restricted to typos/grammar mistakes in hard-coded string UI messages and that type of thing. Good or bad? Thoughts?
So long as it gets marked as 'submitted for QA review' (or however your process works) and someone else in QA has to review and verify it is fixed, then I don't see any real problem. Also assumes (ha!) he sticks to what his proficiency allows him to understand.
-
Gary Wheeler wrote:
"Benefits of a classical education."
You can converse in a dead language. :~
The report of my death was an exaggeration - Mark Twain
Simply Elegant Designs JimmyRopes Designs
I'm on-line therefore I am. JimmyRopesAlong with Latin, they teach Cobol?
-
I hope you don't have any quick fixes in the code that you haven't been allowed to fix properly because no-one will approve a change that doesn't add any visible value and which rely on those hard coded strings
greldak wrote:
which rely on those hard coded strings
Anyone who writes code like that should be sacked.
"The whole idea that carbon dioxide is the main cause of the recent global warming is based on a guess that was proved false by empirical evidence during the 1990s." climate-models-go-cold
-
greldak wrote:
which rely on those hard coded strings
Anyone who writes code like that should be sacked.
"The whole idea that carbon dioxide is the main cause of the recent global warming is based on a guess that was proved false by empirical evidence during the 1990s." climate-models-go-cold
Maybe they were or have since left for other reasons but are you sure that everyone including the aforesaid QA is aware of whether or not such code is present in the codebase and coding appropriately. Bear in mind that even though you are aware of how bad some code may be written you may not be allowed to replace it with a proper implementation unless the change is approved. My point is that even such apparently trivial changes can have unforseen consequences particularly in the case of legacy systems which may not have been well maintained in the past. As such any change needs to be properly controlled and tested. Are you certain that the aforesaid QA guy is following that principle - indeed just by doing so it would appear he is not.
-
Maybe they were or have since left for other reasons but are you sure that everyone including the aforesaid QA is aware of whether or not such code is present in the codebase and coding appropriately. Bear in mind that even though you are aware of how bad some code may be written you may not be allowed to replace it with a proper implementation unless the change is approved. My point is that even such apparently trivial changes can have unforseen consequences particularly in the case of legacy systems which may not have been well maintained in the past. As such any change needs to be properly controlled and tested. Are you certain that the aforesaid QA guy is following that principle - indeed just by doing so it would appear he is not.
We are talking a display string here. That's all.
"The whole idea that carbon dioxide is the main cause of the recent global warming is based on a guess that was proved false by empirical evidence during the 1990s." climate-models-go-cold
-
We are talking a display string here. That's all.
"The whole idea that carbon dioxide is the main cause of the recent global warming is based on a guess that was proved false by empirical evidence during the 1990s." climate-models-go-cold
And I had given an example where even that simple display string changing could change functionality. Just because something appears to be trivial does not mean that it is. It is unlikely that the QA guy will be sufficiently familiar with the code to determine that without spending considerable time reviewing the code. It will be much quicker and safer to have those familiar with the code make such a change. By all means he can identify what he believes needs to be done but he should not be making the change himself.
-
And I had given an example where even that simple display string changing could change functionality. Just because something appears to be trivial does not mean that it is. It is unlikely that the QA guy will be sufficiently familiar with the code to determine that without spending considerable time reviewing the code. It will be much quicker and safer to have those familiar with the code make such a change. By all means he can identify what he believes needs to be done but he should not be making the change himself.
yawn.... Yeah yeah yeah, we all know what can happen, but we are talking about display strings. Name one instance where oy have even seen code that compares or uses in some way "Enter you age" or some such.
"The whole idea that carbon dioxide is the main cause of the recent global warming is based on a guess that was proved false by empirical evidence during the 1990s." climate-models-go-cold
-
yawn.... Yeah yeah yeah, we all know what can happen, but we are talking about display strings. Name one instance where oy have even seen code that compares or uses in some way "Enter you age" or some such.
"The whole idea that carbon dioxide is the main cause of the recent global warming is based on a guess that was proved false by empirical evidence during the 1990s." climate-models-go-cold
I've seen no shortage of cases where business rules have changed to accomodate legislative changes which only impact certain groups and someone in the past has taken the shortcut of hardcoding the changes rather than allowing them to be configurable. I've even come across code in the past which checks the colour of text being displayed far less the actual text. More common would be in the case of contents of comboboxes where the text displayed is checked rather than the value. When you take over a system that 20 years old and been hacked around by as may developers in that time you get to see a lot of horrors.
-
I've seen no shortage of cases where business rules have changed to accomodate legislative changes which only impact certain groups and someone in the past has taken the shortcut of hardcoding the changes rather than allowing them to be configurable. I've even come across code in the past which checks the colour of text being displayed far less the actual text. More common would be in the case of contents of comboboxes where the text displayed is checked rather than the value. When you take over a system that 20 years old and been hacked around by as may developers in that time you get to see a lot of horrors.
SO that's actually a code review failure and nothing to do with QA.
"The whole idea that carbon dioxide is the main cause of the recent global warming is based on a guess that was proved false by empirical evidence during the 1990s." climate-models-go-cold
-
So I'm a dev manager and our QC lead who is gaining proficiency in coding, though is by no means even a junior programmer has taken it upon himself to directly fix some easy bugs. This is certainly a faster way to get things fixed as our dev resources are severely limited. So far it's restricted to typos/grammar mistakes in hard-coded string UI messages and that type of thing. Good or bad? Thoughts?
-
It should be fine if it's only UI messages - if they're kept outside of variables. Such as a messagebox with a fixed caption, or a constant variable. Otherwise you risk coding problems such as "I'll just change the text that's passed to this strcpy()... oops." "I'll just change the text in this char[26]... oops." If the changes they make are under version control the coder could always review them. The thing is, if they do break it, they can just hide behind "I'm not a programmer"... quite legitimately.
SortaCore wrote:
If the changes they make are under version control the coder could always review them
If the changes are not under version control then that is a problem regardless of who is making the change.
SortaCore wrote:
The thing is, if they do break it, they can just hide behind
If they broke it then they broke it. Period. The choice after that is whether they still get a chance to not break it in the future.