Beta test

Or should I say a month of playtesting. A good number of bugs are identified and fixed and publishing a pipeline (to beta test at least) is established. Also, the game UI finally looks like a game UI.

At least that was the state of the game in May. This is going to be a long post, game development is going faster than blogging. So let's begin. Here is the trailer! And information about how to join beta:

It took me a number of attempts to make this video and it shows. I have decent experience in chopping and compressing a video but this was my time composing one out of multiple pieces. The audio was the easiest part, I wrote the text, tried a few recording programs, a few microphones, and read the text the best combination. In principle that is. I had run through multiple drafts of the text because the speech is not a blog post. You have to use simpler sentence structure, watch your breath and rephrase hard to pronounce words. But at least it was quick and easy to iterate. I settled on recording one paragraph (2-3 sentences) multiple times at the time and stitching the best takes in one audio file afterward. Getting video footage was a mess. The game is not integrated with Google Play Games so I couldn't use their screen recording feature, which is otherwise very ergonomic and makes good quality footage. So I had to look for other programs and make do with a barely passable one. Composing a video out of multiple ones was an even bigger mess. I was unable to find a good no-fuss program for it so I had to stretch the capabilities of a tool I worked with before. It again took multiple tries to get desired clips out of recordings because the tool doesn't like making cuts at arbitrary frames and appending multiple such hack job video files often resulted in the glitchy final video. It was the FLV encoder that saved me from non-keyframe cuts hell. And now, now the game is changed so much that I should make another trailer... I dread about it but I'm prepared with better software now and all I really need, besided courage, is to find the time to make the footage.

Ok, so the trailer got uploaded, the link posted here and people came, and then what happened? I'll get to that, after elaborating one little detail. Getting a beta test invite link requires basically having an app published, which means, filling in more forms, figuring out app signing, getting comfortable with the release treadmill, and then wait for a few days to get the app reviewed. The upside of the process is that Google will do some basic app tests for you and the results could be useful for improving the app. Automated tests do have trouble with navigating game map because they don't do touch gestures but they will do everything possible on regular UI widgets and report all sorts of accessibility issues. Also, Google will try running your app on older devices if you declared it should be compatible with them. This test has found (crashed) a few places where the game was using functions unavailable in the old versions of Android.

Ok, so people came, and what happened? When you present a program to the general public, keep in mind that most feedback will not be aligned with your plans for the project and that requests for additions of a similar kind will not be uncommon. Don't get overwhelmed by it, keep calm and try to compromise.

I've limited promoting the game so I got about 10 beta testers. Most of the requests were for more role-playing features. Ones that come to mind are flag emblems, named fleets, and AI personalities. But the most requested change was to make a black background. Since the game kind of already had that in the form of the night mode theme, I made the night mode the default.

At least that was the first step. The second step was to make properly set up application's theme, for both night and day mode and both portrait and landscape orientation. This took the most of the May, mostly for learning which leavers to pull to get the desired effect. But there were also trials and errors in making and setting button and panel background images. Android, especially older OS versions, has very limited support for stretchable images, both in terms of capabilities and documentation. I managed to get things working under imposed constraints but it's not foolproof there are still some quirks remaining I cannot find a solution for.

Other smaller improvements done during the month:

• Fleets can be redirected in flight. Immediate destination (first waypoint) still can't be changed, the rest of the path can be. The idea is that once you send a fleet it is committed to that destination but everything beyond the first destination is a future order and therefore prone to change. 
• Star type and star trait are listed in the star/colony info panel.
• Speed boost technological is effective only between own colonies. The idea is to help speed up ship deployment but to still have an advance warning when being attacked.
• The game now tracks how many qubits you got from each source instead of one total number. This makes it possible to change rank-based rewards in the future and consolidates data for rewards like beating previous survival record.
• Players are first ranked by whether they survived until the end of the game or not and then by score (within alive and eliminated groups). I entertained the idea of having it possible for an eliminated high scoring player to rank above an alive but low scoring one but this would defeat the point of making a big alliance to take out dominating player and just make alive players postpone concluding a finished game until they outscore eliminated players.
• Computer players now gather at least 3 combat ships before attacking. It's a quick and dirty patch for preventing bots from sending ships one at a time in the early game and just losing them without incurring any damage to an enemy with slightly higher defense tech.
• Research points each player gains is not affected by combat outcomes (such as losing a colony). The amount shown in the research screen is guaranteed and technology advancement happens after combat.
• Survival threat ships spawn at random locations outside the map. Quick and dirty partial unstub of this part of the game until I come around implementing it properly.
• Save file can be exported so if the game is crashing in certain situations, you can help by sending me your save.
• Infrastructure for upgrading save files when the game updates, since now there are players other than me.
• Optional ads at the end of the normal game to increased qubit reward.
  

Bug fixes:

• Fleet orders are properly saved.
• Computer player logic starts properly in survival mode.
• Computer player war declaration trigger properly calculates ally strengths so it doesn't change its mind about relation with the strongest player every turn.
• Computer player doesn't crash when sending troop ships.
  
• Colonization news when the whole map is colonized happens only once per game (was repeating itself every turn).
• Attempting colonization on the already colonized world doesn't prematurely end turn processing anymore.
• Selecting an object on the map doesn't change orders given to a previously selected one. Before, selecting an uncolonized star you don't want to colonize would remove colonization order from a previously selected star, and the same happened for stimulation and factory prioritization on own colonized stars.
  
• Star name uniqueness is guaranteed.
• Technologies can advance by multiple levels if there are enough research points.
  

June

 

Most of June was spent improving rendering performance. Yup, even a game as simple as this can have FPS issues. Occasionally at least. Just drawing the map was pretty fast but any map transformation (panning and zooming) would create a lot of short-lived objects, resulting in a performance drop on older phones. Improving the performance was tricky. An initial approach was supposed to be simple, just delegate translation (result of panning) and scale (zoom factor) numbers to the drawing API. But this also scales things you don't want it to: line thickness and text size. For such cases, I still had to make "manual" transformations but I've tried to minimize the work that happens there.

 

Another big thing, simply due to the amount of ground to cover, this month was more info in the game. You can long tap on almost any UI element in the game and get clarification on it. This is akin to right-click in Master of Orion 2 and Master of Magic.

Other smaller improvements done during the month:

• Limited player length to 2 to 100. 
• Changed starting conditions, each player starts with a fully developed colony and one cruiser to start scouting with.
• Perform negotiations with other players by long taping their names on galaxy map info sheets.
• Increased research benefits to x1.2 per level and changed research cost. The first few levels have handpicked costs and later levels grow by factor x1.5.

Bug fixes:

• Diplomatic contacts are checked when the game is created.
• Prevented homeworld placement overlap that was causing players staring eliminated on a crowded map.
  

July

As of yet, July doesn't come with big works like the previous two months so I seized the opportunity to tie up some loose ends and add improve gameplay. One thing I wanted to do for a long time was to add cheats to speed up testing new features. Like being able to see the whole map when working on graphics or jumping straight to survival mode when working on it. Until now I was modifying code to achieve that but that was always running the risk of inadvertently slipping into an official build. And if nothing else, remembering the place I need to hack to get a certain cheat got old pretty fast. Formalizing cheats into app feature neatly solves those problems and it's easy to limit it only to development builds.

Cheats for ending the normal game greatly helped to test the survival mode, which I wanted to align with the design document. Now the replicators don't send their probes before turn 100, if you manage to end the normal mode before it, and their escalation progression counts from that moment rather than from whenever the survival mode was started. This is mostly to prevent players from building up the economy before entering survival mode but also to not overwhelm players who start the mode early. Another major survival mode change is how replicators move, now their fleets attack a colony that is nearest to their spawn location and then moves toward the center of the map, taking time destroying colonies along the way. Once they reach the center, they stay there, I'll have to come up with something more interesting. Also, replicators now have more varied ship types in the fleet and their tech progression is slower.

Survival mode UI was also improved. Contact list now sorts players by population since the score makes no sense in this mode. Population count is also a good indicator of how long a player can withstand a replicator threat. A moment when a player is eliminated is tracked by the game and game over screen for the survival mode ranks the player in order the managed to stay alive.

Other smaller improvements done during the month:

• Ground defense troop growth speed is tied to the number of factories. Previous constant growth was too fast and messed with bombardment a lot.
• Rebalanced bombardment by increasing base bombardment damage and having chemistry reduce received bombard damage. This way early game when you have a few ships, the bombardment is strong enough but doesn't become overpowered later on. 
• The star with ancient megastructure is guarded by high-tech hostile ships. My favorite strategy for getting an advantage over computer players was to colonize central star early. Admittedly that was super unfair and needed to be nerfed.
• Info sheet raised/lowers state is preserved when ending a turn or rotating a phone. It was annoying to me to have the colony info sheet disappear when ending a turn, every turn. Now I only need to figure out how to keep a fleet selected over turns.

Bug fixes:

• The game ends when a human player is eliminated. It would be an interesting feature to see the computer playing against itself, may I'll implement it in the future. Until it's formally coded, there is no point game going on without a human, both normal mode and survival mode.
• Survival can't be started if a human is eliminated.
  
• Fixed crash when 5 player game ends. An interestingly specific corner case :).
  
• Fixed normal game always giving a maximum reward. For some reason, the reward was determined at the start of the game and updated only when loading a game (when a particular object was recreated).