Hi! Thanks for your interest in Hive Project. For later reference, can you please share the corresponding Hive jira ticket for this particular issue? Thanks!
Regards, Laszlo Bodor 浪迹天涯 <1005131...@qq.com.invalid> ezt írta (időpont: 2023. nov. 20., H, 8:51): > Sorry to bother you, I find that this problem has been fixed three years > ago. You can ignore this email,thanks! > > > > 发自我的iPhone > > > ------------------ Original ------------------ > From: It <1005131...@qq.com> > Date: Sat,Nov 18,2023 6:06 PM > To: dev <dev@hive.apache.org> > Subject: Re: Finding HiveBaseResultSet have performance improvements and > wanting to participate the Hive project > > > > Dear Hive team, > I hope this email finds you well. > I am a user of the Hive Open Source project on GitHub and I > have been using it on my company’s project for some time now.Recently, I > noticed that a piece of code in HiveBaseResultSet.java can have > apparent performance improvement with the help of Arthas. However, I sadly > find that there is no open issue can be related to Hive project. So I’m > writing to participate in Hive project. > With the help of Arthas, I can see the whole stack > information of HiveBaseResultSet while getting result from Hive > DataBase, which in our company’s query framework. I found there is nearly > 20% time wasting on String.split() and String.equalsIgnoreCase().So after I > checked the source code with questions, I found the answer on Function > “findColumn()” ! Every time we use result.getXX(String columnName), such as > getObject(String columnName), HiveBaseResultSet will traverse > normalizedColumnNames which is a list used to store the mapping of > columnName and columnIndex, then it gets the index to get the final value. > It wastes too many time on finding the columnName, which includes lowing > cases, traversing, splitting columnName and columnIndex, and > comparing values! So I decided to use a cache to store the mapping of > columnName and columnIndex, which only spends time on storing the mapping > in the first time, and after that, getting index will shorten plenty of > time in searching result, which improve nearly 20% performance on my > project😉. > I was proud of improving my Java skills and really wanted to > make some contribution on Hive project, but unfortunately I noticed that > there is no button to create new issues. Please let me know if there is any > way I can get involved in the Hive project. > Thank you for your time and consideration! And it’s my great > pleasure to discuss code with Hive team. > > > Sincerely, > Followers Xrain_ts > > > > 发自我的iPhone